WO2015145766A1 - Color estimation device, color estimation method, and color estimation program - Google Patents

Color estimation device, color estimation method, and color estimation program Download PDF

Info

Publication number
WO2015145766A1
WO2015145766A1 PCT/JP2014/059303 JP2014059303W WO2015145766A1 WO 2015145766 A1 WO2015145766 A1 WO 2015145766A1 JP 2014059303 W JP2014059303 W JP 2014059303W WO 2015145766 A1 WO2015145766 A1 WO 2015145766A1
Authority
WO
WIPO (PCT)
Prior art keywords
color
product
color name
estimation
pixels
Prior art date
Application number
PCT/JP2014/059303
Other languages
French (fr)
Japanese (ja)
Inventor
宗 益子
陽一 吉本
Original Assignee
楽天株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 楽天株式会社 filed Critical 楽天株式会社
Priority to JP2014527102A priority Critical patent/JP5687806B1/en
Priority to US15/022,243 priority patent/US10692133B2/en
Priority to PCT/JP2014/059303 priority patent/WO2015145766A1/en
Publication of WO2015145766A1 publication Critical patent/WO2015145766A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0641—Shopping interfaces
    • G06Q30/0643—Graphical representation of items or shoppers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/56—Extraction of image or video features relating to colour

Definitions

  • One aspect of the present invention relates to a color estimation device, a color estimation method, and a color estimation program.
  • Patent Document 1 a technique for specifying a color having a large number of pixels contained in an image showing a product as a main color of the product is known (see, for example, Patent Document 1).
  • the background color may be erroneously specified as the product color name.
  • the product color name considering that there is a high possibility that the product is placed in the center of the image, it may be possible to calculate the number of pixels by weighting the pixel located in the center of the image. If different colors are arranged in the center of the image, an incorrect color name may be specified.
  • an object of one aspect of the present invention is to accurately estimate the color of a product shown in a product image including a plurality of colors.
  • a color estimation device includes a plurality of color text information representing a color name from product information associated with a product image in which the product is represented. Extracting means for extracting as, a pixel value set for each color name candidate, and an estimating means for estimating the color of the product represented in the product image based on the pixel value of the pixel included in the product image; Output means for outputting the color estimated by the estimation means.
  • a color estimation method is a color estimation method executed by a computer, and a plurality of color text information representing a color name is obtained from product information associated with a product image in which the product is represented.
  • the color of the product represented in the product image is estimated based on the extraction step for extracting the color name candidate, the pixel value set for each color name candidate, and the pixel value of the pixel included in the product image And an output step for outputting the color estimated in the estimation step.
  • the color estimation program is an extraction function for extracting, from a product information associated with a product image representing a product, a plurality of color text information representing a color name as color name candidates for the product. And an estimation function that estimates the color of the product represented in the product image based on the pixel value set for each color name candidate and the pixel value of the pixel included in the product image. An output function for outputting the selected color.
  • color name candidates are extracted from the product information including text indicating the color of the product represented in the product image, and based on the pixel value and the number of pixels included in the product image, the color of the product Therefore, it is suppressed that an incorrect color is estimated as the color name of the product, and accurate color estimation is realized.
  • the estimation means estimates the color of the product represented in the product image from the color name candidates extracted by the extraction means.
  • the estimation unit calculates, for each color name candidate, the number of product image pixels included in a color range that is a range of pixel values set for the color name candidate.
  • the color name candidate having the largest number of pixels is estimated as the color of the product represented in the product image.
  • the color name candidate having the largest number of pixels included in the color range set for each color name candidate is estimated as the color of the product
  • the color most likely to represent the color of the product Candidate names can be output as product colors.
  • the estimating unit weights the second pixel located closer to the center of the product image than the first pixel included in the product image to a weight greater than the first pixel. The number of pixels for each color name candidate is calculated.
  • the number of pixels for each color name candidate is calculated. Therefore, it is possible to estimate the color with higher accuracy.
  • the estimation unit sets the color range for each color name candidate so as not to overlap based on the pixel value set for the color name candidate.
  • the color ranges set for the color name candidates are set without duplication, even if the extracted color name candidates have close pixel values, the color name candidates are appropriately displayed. It is possible to calculate the number of pixels for each.
  • the estimation unit clusters the pixels of the product image based on the pixel values, and when the median value of the group is included in the color range of the color name candidate, The number of pixels for each color name candidate is calculated by using the number as the number of pixels included in the color range of the color name candidate.
  • the region having the color that is the color of the product occupies a certain range, so according to this aspect, the color name candidates are associated with each group of pixels clustered by pixel values, Color name candidates are associated with each similar color region, and the number of pixels is calculated. As a result, it is possible to efficiently estimate the color with high accuracy.
  • the estimation unit includes a color name candidate having the largest number of calculated pixels, and a color name having a pixel number that is equal to or less than a predetermined number of difference between the color name candidate and the calculated pixel number. Candidates are estimated as the colors of the plurality of products represented in the product image.
  • the estimation means estimates a color for each product image when a plurality of product images are associated with one product.
  • the color of the product represented in each product image can be estimated.
  • the estimation unit when a plurality of product images are associated with one product, the estimation unit is set for one color name candidate among the plurality of color name candidates.
  • the number of product image pixels included in the color range that is the range of pixel values is calculated for each of the plurality of product images, and the color of the product image with the largest number of pixels is estimated as one color name candidate. .
  • the number of pixels having a pixel value corresponding to the color name of each product image is calculated, and one color name candidate is selected as the color of the product image having the largest number of pixels. Therefore, even when the extracted plurality of color name candidates have similar pixel values, it is possible to accurately estimate the color.
  • the estimation unit when a plurality of product images are associated with one product, has a color range that is a pixel value range set for the color name candidate.
  • the number of pixels of one product image included is calculated for each color name candidate, and the color name candidate with the largest number of calculated pixels is estimated as the color of the product represented in one product image, and multiple colors
  • the range of pixel values set for one color name candidate among a plurality of color name candidates when there is a color name candidate that is not estimated as the product color of any product image Is calculated for each of the plurality of product images, and the color of the product image having the largest number of pixels is estimated as one color name candidate.
  • the color name candidate having the largest number of pixels included in the color range set for each color name candidate is estimated as the color of the product represented in the one product image.
  • a pixel having a pixel value corresponding to the color name of each product image is estimated as the color name of the product image with the largest number of pixels.
  • the extraction unit may extract the color text information by searching the field when the product information includes a field for accepting designation of the product color from the user and the color text information. Extract as
  • color name candidates can be extracted appropriately.
  • the output unit stores the estimated color in association with the product image of the product information storage unit that stores the product information and the product image in association with each other.
  • the product image can be processed based on the color.
  • a color estimation apparatus refers to a product information storage unit in response to a search request from a user including specification of a product color, and includes a product image associated with a specified color.
  • a search means for returning the result to the user is further provided.
  • the product image having the color can be provided to the user in response to the search request for the product image specifying the color from the user.
  • a plurality of product images are associated with one product, and one product image of the plurality of product images presents the product.
  • the search means specifies when a search request from a user including a color specification different from the color associated with the representative image is received.
  • a search result including the product image associated with the color as a representative image is returned to the user.
  • FIG. 9A is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “red” and “blue”.
  • FIG. 9B is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “blue”, “green”, and “white”. It is a figure which shows the example of the goods information storage part in which the estimated color name was memorize
  • FIG. 4 is a diagram schematically illustrating two-dimensional color ranges of color name candidates “red”, “dark green”, and “yellow green” in a color space. It is a flowchart which shows the example of the processing content of the color estimation method in a color estimation apparatus.
  • FIG. 17A is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P51.
  • FIG. 17B is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P52.
  • FIG. 17C is a diagram illustrating an example of the number of pixels that are calculated for each of the color name candidates “black” and “dark blue” and each pixel of the product images P51 and P52 is included in the color range. It is a flowchart which shows the content of the further another example of the estimation process of the color name by an estimation part.
  • FIG. 1 is a block diagram showing a functional configuration of the color estimation apparatus 1 according to the present embodiment.
  • the color estimation device 1 of the present embodiment is a device that estimates the color name of a product represented in a product image.
  • the color estimation device 1 is configured as a server, for example, but may be configured integrally with the electronic commerce server 3 configuring the electronic commerce site.
  • the electronic commerce server 3 is a server that controls commerce via the Internet, and returns a search result in response to, for example, a search request for a product from a user.
  • the color estimation apparatus 1 functionally includes an extraction unit 11 (extraction unit), an estimation unit 12 (estimation unit), and an output unit 13 (output unit).
  • Each functional unit of the color estimation apparatus 1 can access storage means such as a product information storage unit 21 (product information storage unit) and a color information storage unit 22.
  • FIG. 2 is a hardware configuration diagram of the color estimation apparatus 1.
  • the color estimation apparatus 1 physically includes a CPU 101 configured by a processor, a main storage device 102 configured by a memory such as a RAM and a ROM, an auxiliary storage device 103 configured by a hard disk, and the like.
  • the computer system includes a communication control device 104 constituted by a network card or the like, an input device 105 such as a keyboard and mouse as input devices, an output device 106 such as a display, and the like.
  • Each function shown in FIG. 1 is a communication control device under the control of the CPU 101 by reading predetermined computer software (color estimation program) on the hardware such as the CPU 101 and the main storage device 102 shown in FIG. 104, the input device 105, and the output device 106 are operated, and data is read and written in the main storage device 102 and the auxiliary storage device 103. Data and databases necessary for processing are stored in the main storage device 102 and the auxiliary storage device 103.
  • the extraction unit 11 is a part that extracts a plurality of color text information representing color names as product color name candidates from the product information associated with the product image representing the product.
  • the product information includes text information regarding the product to be presented to the user.
  • the text information related to the product is, for example, an explanatory note related to the product, a color name accompanied with a column for accepting designation of the color of the product on the product page.
  • the product information storage unit 21 stores information related to products used for electronic commerce, and stores product information and product images in association with product IDs for identifying products.
  • FIG. 3 is a diagram illustrating a configuration of the product information storage unit 21 and an example of stored data. As illustrated in FIG. 3, for example, the product information storage unit 21 stores product information T1 and a product image P1 in association with the product ID (M1).
  • the product information T1 includes, as its content, text information to be presented to the user that “there are red and blue colors”.
  • the product information storage unit 21 stores product information T2 and a product image P2 in association with the product ID (M2).
  • the product information T2 includes text information “... There are three colors of blue, green and white,...
  • the product information storage unit 21 stores product information T3 and a plurality of product images P31, P32, and P33 in association with the product ID (M3).
  • the product information T3 includes, as its contents, color text information “blue”, “red”, and “yellow” with a column for accepting designation of the product color on the product page. These fields and color text information are for presentation to the user.
  • the product image P31 among the product images P31, P32, and P33 is set as a representative image when the product with the product ID (M3) is displayed as a product page.
  • FIG. 4 is a diagram illustrating an example of the product page of the product ID (M1) displayed based on the information stored in the product information storage unit 21.
  • the product page with the product ID (M1) includes product information T1 including a product image P1 representing the product and an explanatory text “There are red and blue colors”. That is, the product information T1 includes color text information “red” and “blue” representing the color name. There is a high possibility that the color text information corresponds to the color of the product shown in the product image P1.
  • FIG. 5 is a diagram illustrating an example of the product page of the product ID (M2) displayed based on the information stored in the product information storage unit 21.
  • the product page of the product ID (M2) is composed of a product image P2 representing the product and a description regarding the product “... There are three colors of blue, green, and white,.
  • Product information T2 is included. That is, the product information T2 includes color text information “blue”, “green”, and “white” representing the color name. There is a high possibility that the color text information corresponds to the color of the product shown in the product image P2.
  • FIG. 6 is a diagram illustrating an example of the product page of the product ID (M3) displayed based on the information stored in the product information storage unit 21.
  • the product page with the product ID (M3) includes a product image P31 set as a representative image.
  • the product images P32 and P33 other than the representative image may be displayed in a thumbnail format.
  • the product page includes a plurality of columns for receiving designation of product colors and product information T3 including color text information “blue”, “red”, and “yellow” corresponding to each column. That is, there is a high possibility that the color of the product shown in the product image P31 is any one of “blue”, “red”, and “yellow”.
  • the extraction unit 11 extracts color text information as color name candidates for the product from the product information shown in FIGS. 3, 4, 5, and 6.
  • the extraction unit 11 can extract color text information by referring to a color name in the color information storage unit 22 described later as a dictionary. Specifically, the extraction unit 11 extracts the color text information “red” and “blue” from the product information T1 as color name candidates for the product with the product ID (M1). Further, the extraction unit 11 uses the color text information “blue”, “green”, and “white” from the plurality of fields for receiving the designation of the product color from the product information T2, and the product color of the product ID (M2). Extract as name candidates. Further, the extraction unit 11 extracts the color text information “blue”, “red”, and “yellow” from the product information T3 as product color name candidates of the product ID (M3).
  • the color text information extracted from the product information may include variations such as “black” and “black” even if the color names indicate the same color.
  • the extraction unit 11 may normalize the extracted color text information and set it as a color name candidate. That is, when the color text information “black” and “black” is extracted from the product information, the extraction unit 11 normalizes these color text information to “black” and sets them as color name candidates.
  • the estimation unit 12 is a unit that estimates the color of the product represented in the product image based on the pixel value set for the color name candidate extracted by the extraction unit 11 and the pixel value of the pixel included in the product image. It is.
  • the estimation unit 12 estimates the color name of the product represented in the product image from the color name candidates extracted by the extraction unit 11. That is, the estimating unit 12 calculates the number of pixels of the product image included in the color range that is the range of pixel values set for the color name candidate for each color name candidate, and the calculated number of pixels is the largest.
  • the color name candidate is estimated as the color name of the product shown in the product image.
  • the estimation unit 12 refers to the color information storage unit 22 and acquires a color range corresponding to a color name candidate.
  • FIG. 7 is a diagram illustrating a configuration of the color information storage unit 22 and an example of stored data.
  • the color information storage unit 22 stores pixel values set for each color name. Specifically, the color information storage unit 22 stores a color range set for each color name. Further, the color information storage unit 22 may further store a median value for each color name.
  • the pixel value is expressed in the HSL color space, but it is not excluded that it is expressed in other parameters (for example, HSV space, RGB space, etc.).
  • the color range of the color name “red” is (H 1L to H 1H , S 1L to S 1H , L 1L to L 1H ), and the median is (H 1M , S 1M , L 1M ). .
  • FIG. 8 is a diagram schematically showing the HSL color space.
  • the estimation unit 12 acquires the color name candidates “red” and “blue” extracted from the product information T1 from the color information storage unit 22.
  • the color range of the color name “red” is indicated by the color range CR.
  • the color range of the color name “blue” is indicated by the color range CB.
  • the estimation unit 12 calculates the number of pixels of the product image P1 included in each of the color range CR of the color name candidate “red” and the color range CB of the color name candidate “blue”.
  • FIG. 8 is a diagram schematically showing the HSL color space.
  • the estimation unit 12 acquires the color name candidates “red” and “blue” extracted from the product information T1 from the color information storage unit 22.
  • the color range of the color name “red” is indicated by the color range CR.
  • the color range of the color name “blue” is indicated by the color range CB.
  • the estimation unit 12 calculates the number of pixels of the
  • FIG. 9A is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “red” and “blue”.
  • the estimation unit 12 estimates the color name of the product represented in the product image P1 as “red”. In this way, the estimation unit 12 can estimate the color name candidate that is most likely to represent the color of the product as the color name of the product.
  • the estimation unit 12 weights the second pixel closer to the center of the product image than the first pixel included in the product image, more than the first pixel.
  • the number of pixels for each color name candidate may be calculated.
  • the estimation unit 12 may calculate the number of pixels by weighting pixels located within a predetermined range including the central part of the product image, or in proportion to the distance from the central part. The number of pixels may be calculated by weighting. As a result, more weight is given to the pixels near the central portion where there is a high possibility that a more appropriate color is arranged as the color name of the product in the product image, and the number of pixels for each color name candidate is calculated. It is possible to estimate a color name with high accuracy.
  • the estimation unit 12 may exclude pixels in a region other than the region in which the product is represented in the product image from the target for calculating the number of pixels.
  • a region other than the region where the product is represented in the product image can be extracted by a known image processing technique.
  • the estimation unit 12 may exclude the pixels in the background area from the target for calculating the number of pixels.
  • the estimation unit 12 For the product with the product ID (M2) (see FIGS. 3 and 5), the estimation unit 12 stores the color range of the color name candidates “blue”, “green”, and “white” extracted from the product information T2 as color information. Obtained from the unit 22. Then, the estimation unit 12 calculates the number of pixels of the product image P2 included in the color ranges of the color name candidates “blue”, “green”, and “white”.
  • FIG. 9B is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “blue”, “green”, and “white”. As shown in FIG.
  • the number of pixels whose pixel value is included in the color range of the color name candidate “blue” is 180, and the color range of the color name candidate “green”
  • the number of pixels including the pixel value is 175, and the number of pixels including the pixel value in the color range of the color name candidate “white” is 45.
  • the estimation unit 12 displays a color name candidate having the largest number of calculated pixels and a color name candidate having a number of pixels whose difference between the color name candidate and the calculated number of pixels is a predetermined number or less in the product image. It can be estimated as the color name of each of a plurality of products. Specifically, in the example of the product with the product ID (M2), the predetermined number regarding the difference in the number of pixels for estimating the plurality of color name candidates as the color names of the plurality of products represented in the product image is 10. Explained.
  • the color name candidate having the largest number of pixels of the product image P2 included in each color range is “blue” (180 pixels). Since the number of pixels included in the color range of “green” is 175 with respect to this number of pixels and the difference is 10 or less, the estimation unit 12 assigns “blue” and “green” to the product image P2. Estimated as the color name of each of the multiple products represented in. By such an estimation process, it is possible to estimate the color name of each product even when a plurality of products with different colors are represented in one product image.
  • the estimation unit 12 For the product with the product ID (M3) (see FIGS. 3 and 6), the estimation unit 12 stores the color range of the color name candidates “blue”, “red”, and “yellow” extracted from the product information T3 as color information. Obtained from the unit 22.
  • the estimation unit 12 estimates a color name for each product image. Specifically, because the product information T3 and the plurality of product images P31, P32, and P33 are associated with the product with the product ID (M3), the estimation unit 12 determines each of the product images P31, P32, and P33.
  • the number of pixels included in each color range of the color name candidates “blue”, “red”, and “yellow” is calculated, and each product image
  • the color name of the product represented by P31, P32, and P33 is estimated.
  • the description of the estimation process indicating the specific number of pixels is omitted.
  • the estimation unit 12 calculates the number of pixels of the product image included in the color range for each color name candidate, but instead of the number of pixels, the product image
  • the ratio of the pixels included in the color range occupied by the pixels may be calculated and estimated as the color name of the product displayed in the product image.
  • the estimation unit 12 estimates the color name of the product image of the product image based on the pixel value of the color name candidate extracted by the extraction unit 11. It is good also as estimating the color specified as. That is, as the color to be estimated, a color name that is text information may be estimated, or a pixel value indicating a displayed color may be estimated.
  • the output unit 13 is a part that outputs the color name estimated by the estimation unit 12. Specifically, the output unit 13 stores the estimated color name in association with the product image in the product information storage unit 21 that stores the product information and the product image in association with each other.
  • FIG. 10 is a diagram illustrating an example of the product information storage unit 21 in which the estimated color names are stored. As illustrated in FIG. 10, the output unit 13 stores the estimated color name “red” in association with the product image P1. Further, the output unit 13 stores the estimated color names “blue” and “green” in association with the product image P2. Further, the output unit 13 stores the estimated color names “blue”, “red”, and “yellow” in association with the product images P31, P32, and P33, respectively. As described above, since the color name is associated with the product image, the product image based on the color name can be processed. For example, a product image can be extracted by specifying a color name.
  • FIG. 11A is a diagram schematically illustrating two-dimensional color ranges of candidate color names “red”, “dark green”, and “yellow green” in the color space. As shown in FIG.
  • the red color range CR and the dark green color range CDG 1 and the yellow green color range CYG 1 are separated from each other, but “dark green” and “yellow green” are Since the pixel values shown are close to each other, the dark green color range CDG 1 and the yellow green color range CYG 1 overlap.
  • the estimation unit 12 can set the color range for each color name candidate so as not to overlap based on the pixel value set for the color name candidate. Specifically, for example, the estimation unit 12 acquires the median value of the color range of “dark green” and “yellow green” from the color information storage unit 22, and uses half the distance between the median values as the color. The radius of the range.
  • FIG. 11B is a diagram schematically illustrating two-dimensionally the color ranges of “red”, “dark green”, and “yellow green” set so as not to overlap. As shown in FIG. 11B, the overlap between the dark green color range CDG 2 and the yellow green color range CYG 2 is eliminated. By setting the color range in this way, the color ranges set for the color name candidates are set without duplication, so even if the extracted color name candidates have close pixel values, It is possible to appropriately calculate the number of pixels for each color name candidate.
  • the estimation unit 12 clusters the pixels of the product image based on the pixel values, and when the median value of the clustered group is included in the color range of the color name candidate, the pixels included in the group
  • the number of pixels for each color name candidate may be calculated using the number of pixels as the number of pixels included in the color range of the color name candidate.
  • color name candidates are associated with each group of pixels clustered according to pixel values, so that color name candidates are associated with each similar color region, and the number of pixels is calculated. Will be done. This makes it possible to estimate a color name with high accuracy and efficiency.
  • the color estimation device 1 may be configured integrally with the electronic commerce server 3 constituting the electronic commerce site (see FIG. 1).
  • the color estimation apparatus 1 refers to the product information storage unit 21 in response to a search request from the user including specification of the color name of the product, and the product associated with the specified color name. It is good also as providing the search part 31 which returns the search result containing an image to a user.
  • search unit 31 when a search request for a product including designation of a product color name (for example, “green”) is received from a user terminal device (not shown) via the network, the search unit 31 Referring to the product information storage unit 21 shown in FIG. 10, information related to the product with the product ID (M2) including the product image P2 associated with the color name “green” is returned as a search result to the user's terminal device.
  • a product color name for example, “green”
  • a search request including a keyword “shirt” corresponding to the product with the product ID (M3) and a color name “red” different from the color name “blue” associated with the representative image of the product ID (M3) is received.
  • the search unit 31 may return a search result including the product image P32 associated with the designated color name “red” as a representative image to the user. That is, with reference to FIG. 6, the search unit 31 returns a product page that displays the product image P32 as a representative image instead of the product image P31 to the user's terminal device.
  • an image having a color name different from the color name related to the search request is set in advance as a representative image of the product in response to a search request for the product accompanied by designation of the color name from the user. Even if it exists, the product image of the color name which concerns on a search request can be provided to a user.
  • FIG. 12 is a flowchart showing an example of processing contents of the color estimation method in the color estimation apparatus 1 shown in FIG.
  • the extraction unit 11 extracts a plurality of color text information representing color names as product color name candidates from the product information associated with the product image (S1).
  • the estimation unit 12 sets the product value represented in the product image based on the pixel value set based on the color name candidate extracted by the extraction unit 11 in step S1 and the pixel value of the pixel included in the product image.
  • a color name is estimated from among color name candidates.
  • the estimation unit 12 calculates the number of pixels of the product image included in the color range set for the color name candidate for each color name candidate (S2). Then, the estimation unit 12 estimates the color name candidate having the largest number of calculated pixels as the color name of the product represented in the product image (S3). Then, the output unit 13 outputs the color name estimated by the estimation unit 12 (S4).
  • the color estimation program 1p includes a main module m10, an extraction module m11, an estimation module m12, and an output module m13.
  • the color estimation program 1p may further include a search module (not shown).
  • the main module m10 is a part that comprehensively controls the color estimation process.
  • the functions realized by executing the extraction module m11, the estimation module m12, the output module m13, and the search module are respectively the extraction unit 11, the estimation unit 12, the output unit 13, and the search unit of the color estimation apparatus 1 shown in FIG. This is the same as the function 31.
  • the color estimation program 1p is provided by a storage medium 1d such as a CD-ROM, a DVD-ROM, or a semiconductor memory, for example.
  • the color estimation program 1p may be provided via a communication network as a computer data signal superimposed on a carrier wave.
  • color name candidates are extracted from product information including text indicating the color of the product represented in the product image. Since the color name of the product is estimated from the color name candidates based on the pixel value and the number of pixels included in the product image, it is possible to prevent an incorrect color name from being estimated as the color name of the product, A good color name estimation is realized.
  • FIG. 14A is a diagram illustrating an example of data stored in the product information storage unit 21.
  • product information T4 and product images P41, P42, and P43 are stored in association with the product ID (M4).
  • the product information T4 includes, as its contents, text information for presenting to the user that “You can choose from three colors, dark brown, dark blue, and black.”
  • the extraction unit 11 extracts color name candidates “dark brown”, “dark blue”, and “black” from the product information T4.
  • the estimation unit 12 sets one color name candidate among the plurality of color name candidates.
  • the number of pixels of the product image included in the determined color range is calculated for each of the plurality of product images, and it is estimated that the color name of the product image having the largest number of pixels is one color name candidate.
  • the estimation unit 12 calculates the number of product image pixels included in the color range set for one color name candidate “dark brown” for each of the product images P41, P42, and P43. Similarly, the estimation unit 12 calculates the number of product image pixels included in the color range set for the color name candidate “dark blue” for each of the product images P41, P42, and P43, and the color name candidate “black”. The number of pixels of the product image included in the color range set for “” is calculated for each of the product images P41, P42, and P43.
  • FIG. 15 illustrates an example of the number of pixels in which the pixels of the product images P41, P42, and P43 are included in the color range, calculated for each of the color name candidates “dark brown”, “dark blue”, and “black”.
  • the estimation unit 12 assigns the color name of the product represented in the product image P41 to “dark brown”. ".
  • the numbers of pixels of the product images P41, P42, and P43 included in the color range of the color name candidate “dark blue” are 130, 200, and 140, respectively. Accordingly, since the product image including the most pixels included in the color range of the color name candidate “dark blue” is the product image P42, the estimation unit 12 sets the color name of the product represented in the product image P42 to “dark blue”. ”.
  • the numbers of pixels of the product images P41, P42, and P43 included in the color range of the color name candidate “black” are 90, 80, and 130, respectively. Therefore, since the product image including the most pixels included in the color range of the color name candidate “black” is the product image P43, the estimation unit 12 sets the color name of the product represented in the product image P43 to “black”. presume.
  • the output unit 13 uses the estimated color names “dark brown”, “dark blue”, and “black” as product images P41, P42 and P43 are stored in association with each other.
  • the color name when the color name is estimated by the process of calculating the number of pixels included in the color range of each color name candidate for each product image, it is included in dark brown among the pixels included in the product image P43. Therefore, the product name P43 is erroneously estimated to be dark brown.
  • the number of pixels of the pixel value corresponding to the color name of each product image is calculated, and the color name of the product image with the largest number of pixels is one. Therefore, even if the extracted color name candidates have similar pixel values, it is possible to estimate the color name with high accuracy.
  • FIG. 16A is a diagram illustrating an example of data stored in the product information storage unit 21.
  • product information T5 and product images P51 and P52 are stored in association with the product ID (M5).
  • the merchandise information T5 includes text information to be presented to the user, “There are black and dark blue.”
  • the extraction unit 11 extracts color name candidates “black” and “dark blue” from the product information T5.
  • the estimation unit 12 calculates the number of pixels of the product image P51 included in the color range set for each color name candidate for each color name candidate.
  • FIG. 17A is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P51.
  • the estimation unit 12 estimates “black” having the largest number of calculated pixels as the color name of the product shown in the product image P51.
  • the estimation unit 12 calculates the number of pixels of the product image P52 included in the color range set for each color name candidate for each color name candidate.
  • FIG. 17B is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P52.
  • the estimation unit 12 estimates “black” having the largest number of calculated pixels as the color name of the product represented in the product image P52.
  • the output unit 13 outputs the color name “black” estimated for the product image P51 and the color name “black” estimated for the product image P52 to the product information storage unit.
  • the product images P51 and P52 of 21 are stored in association with each other.
  • the color name publication “Dark Blue” is the color name candidate “Dark Blue”, although a plurality of color name candidates “Black” and “Dark Blue” are extracted by the extraction unit 11. Since it is not estimated as the color name of the product of any product image, there is a high possibility that there is an error in the estimation of the color name.
  • the estimation unit 12 is included in the color range set for one color name candidate among the plurality of color name candidates, as in the example described with reference to FIGS. 14 and 15.
  • the number of pixels of the product image to be obtained is calculated for each of the plurality of product images, and it is estimated that the color name of the product image having the largest number of pixels is one color name candidate.
  • the estimation unit 12 calculates the number of product image pixels included in the color range set for the color name candidate “black” for each of the product images P51 and P52. Similarly, the estimation unit 12 calculates the number of product image pixels included in the color range set for the color name candidate “dark blue” for each of the product images P51 and P52.
  • FIG. 17C is a diagram illustrating an example of the number of pixels that are calculated for each of the color name candidates “black” and “dark blue” and each pixel of the product images P51 and P52 is included in the color range.
  • the number of pixels of the product images P51 and P52 included in the color range of the color name candidate “black” is 150 and 140, respectively. Accordingly, since the product image including the most pixels included in the color range of the color name candidate “black” is the product image P51, the estimation unit 12 sets the color name of the product represented in the product image P51 to “black”. presume. Further, the numbers of pixels of the product images P51 and P52 included in the color range of the color name candidate “dark blue” are 100 and 120, respectively. Therefore, since the product image including the most pixels included in the color range of the color name candidate “dark blue” is the product image P52, the estimation unit 12 sets the color name of the product represented in the product image P52 to “dark blue”. ". Then, as illustrated in FIG. 16C, the output unit 13 associates the estimated color names “black” and “dark blue” with the product images P51 and P52 of the product information storage unit 21, respectively. Remember me.
  • FIG. 18 is a flowchart showing the contents of still another example of color name estimation processing by the estimation unit 12.
  • the extraction unit 11 extracts a plurality of color text information representing color names as product color name candidates from the product information associated with the product image (S11).
  • the estimation unit 12 calculates, for each color name candidate, the number of pixels of one product image included in the color range set for the color name candidate (S12). Then, the estimation unit 12 estimates the color name candidate having the largest number of calculated pixels as the color name of the product represented in one product image (S13).
  • the estimation unit 12 determines whether there is a color name candidate that has not been estimated as the product color name of any product image among the plurality of color name candidates (S14). If it is determined that there is a color name candidate that has not been estimated as the product color name of any product image, the processing procedure proceeds to step S15. On the other hand, if it is not determined that there is a color name candidate that has not been estimated as the color name of the product of any product image, the processing procedure proceeds to step S17.
  • step S15 the estimation unit 12 calculates the number of pixels of the product image included in the color range set for one color name candidate among the plurality of color name candidates for each of the plurality of product images. (S15). And the estimation part 12 estimates that the color name of the product image with the largest number of pixels is one color name candidate (S16). The output unit 13 outputs the estimated color name (S17).
  • the color name candidate having the largest number of pixels included in the color range set for each color name candidate is the color of the product represented in the one product image.
  • the color of each product image for one color name candidate The number of pixels having the pixel value corresponding to the name is calculated, and one color name candidate is estimated as the color name of the product image having the largest number of pixels.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Library & Information Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Databases & Information Systems (AREA)
  • Multimedia (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Image Analysis (AREA)

Abstract

This color estimation device is provided with: an extraction means for extracting, from product information that is associated with a product image representing a product, said information including product-related text information to be presented to a user, a plurality of pieces of color text information representing a color name as color name candidates for the product; an estimation means for estimating the color of a product represented by a product image on the basis of a pixel value, which is set on the basis of the color name candidates, and the pixel value of a pixel included in the product image; and an output means for outputting the color estimated by the estimation means.

Description

色推定装置、色推定方法及び色推定プログラムColor estimation apparatus, color estimation method, and color estimation program
　本発明の一側面は、色推定装置、色推定方法及び色推定プログラムに関する。 One aspect of the present invention relates to a color estimation device, a color estimation method, and a color estimation program.
　従来、商品が表された画像に含まれるピクセル数が多い色を商品の主要な色として特定する技術が知られている（例えば、特許文献１参照）。 Conventionally, a technique for specifying a color having a large number of pixels contained in an image showing a product as a main color of the product is known (see, for example, Patent Document 1).
特開２０１１－５２０２０３号公報JP 2011-520203 A
　従来の技術では、例えば商品の画像に背景の領域が多く含まれると、背景の色が商品の色名として誤って特定される可能性がある。また、商品が画像の中央に配置される可能性が高いことに鑑みて、画像の中央部に位置するピクセルに重み付けをしてピクセル数を算出することが考えられるが、その商品の色とは異なる色が画像中央部に配置されていた場合、誤った色名が特定される可能性がある。 In the conventional technology, for example, if a product image includes many background regions, the background color may be erroneously specified as the product color name. In addition, considering that there is a high possibility that the product is placed in the center of the image, it may be possible to calculate the number of pixels by weighting the pixel located in the center of the image. If different colors are arranged in the center of the image, an incorrect color name may be specified.
　そこで本発明の一側面は、複数の色が含まれる商品画像に表されている商品の色を精度良く推定することを目的とする。 Therefore, an object of one aspect of the present invention is to accurately estimate the color of a product shown in a product image including a plurality of colors.
　上記課題を解決するために、本発明の一形態に係る色推定装置は、商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を商品の色名候補として抽出する抽出手段と、それぞれの色名候補に対して設定されたピクセル値と、商品画像に含まれるピクセルのピクセル値とに基づき、商品画像に表された商品の色を推定する推定手段と、推定手段により推定された色を出力する出力手段と、を備える。 In order to solve the above problem, a color estimation device according to an aspect of the present invention includes a plurality of color text information representing a color name from product information associated with a product image in which the product is represented. Extracting means for extracting as, a pixel value set for each color name candidate, and an estimating means for estimating the color of the product represented in the product image based on the pixel value of the pixel included in the product image; Output means for outputting the color estimated by the estimation means.
　本発明の一形態に係る色推定方法は、コンピュータにより実行される色推定方法であって、商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を商品の色名候補として抽出する抽出ステップと、それぞれの色名候補に対して設定されたピクセル値と、商品画像に含まれるピクセルのピクセル値とに基づき、商品画像に表された商品の色を推定する推定ステップと、推定ステップにおいて推定された色を出力する出力ステップと、を有する。 A color estimation method according to an aspect of the present invention is a color estimation method executed by a computer, and a plurality of color text information representing a color name is obtained from product information associated with a product image in which the product is represented. The color of the product represented in the product image is estimated based on the extraction step for extracting the color name candidate, the pixel value set for each color name candidate, and the pixel value of the pixel included in the product image And an output step for outputting the color estimated in the estimation step.
　本発明の一形態に係る色推定プログラムは、コンピュータに、商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を商品の色名候補として抽出する抽出機能と、それぞれの色名候補に対して設定されたピクセル値と、商品画像に含まれるピクセルのピクセル値とに基づき、商品画像に表された商品の色を推定する推定機能と、推定機能により推定された色を出力する出力機能と、を実現させる。 The color estimation program according to an aspect of the present invention is an extraction function for extracting, from a product information associated with a product image representing a product, a plurality of color text information representing a color name as color name candidates for the product. And an estimation function that estimates the color of the product represented in the product image based on the pixel value set for each color name candidate and the pixel value of the pixel included in the product image. An output function for outputting the selected color.
　上記側面によれば、商品画像に表された商品の色を示すテキストが含まれている商品情報から色名候補が抽出され、商品画像に含まれるピクセルのピクセル値及び数に基づき、商品の色が推定されるので、誤った色が商品の色名として推定されることが抑制され、精度の良い色の推定が実現される。 According to the above aspect, color name candidates are extracted from the product information including text indicating the color of the product represented in the product image, and based on the pixel value and the number of pixels included in the product image, the color of the product Therefore, it is suppressed that an incorrect color is estimated as the color name of the product, and accurate color estimation is realized.
　別の側面に係る色推定装置では、推定手段は、商品画像に表された商品の色を前記抽出手段により抽出された色名候補の中から推定する。 In the color estimation apparatus according to another aspect, the estimation means estimates the color of the product represented in the product image from the color name candidates extracted by the extraction means.
　上記側面によれば、色名候補の中から商品の色が推定されるので、誤った色名が商品の色として推定されることが抑制される。 According to the above aspect, since the color of the product is estimated from the color name candidates, it is suppressed that an incorrect color name is estimated as the color of the product.
　別の側面に係る色推定装置では、推定手段は、色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる商品画像のピクセルの数を色名候補ごとに算出し、算出されたピクセル数が最も多い色名候補を、商品画像に表された商品の色として推定する。 In the color estimation device according to another aspect, the estimation unit calculates, for each color name candidate, the number of product image pixels included in a color range that is a range of pixel values set for the color name candidate. The color name candidate having the largest number of pixels is estimated as the color of the product represented in the product image.
　この側面によれば、色名候補ごとに設定された色範囲に含まれるピクセル数が最も多い色名候補が商品の色として推定されるので、商品の色を表している可能性が最も高い色名候補を商品の色として出力できる。 According to this aspect, since the color name candidate having the largest number of pixels included in the color range set for each color name candidate is estimated as the color of the product, the color most likely to represent the color of the product Candidate names can be output as product colors.
　さらに別の側面に係る色推定装置では、推定手段は、商品画像に含まれる第１のピクセルより商品画像の中央部に近い位置にある第２のピクセルに対して、第１のピクセルより大きい重み付けをして、色名候補ごとのピクセル数の算出を行う。 In the color estimation device according to another aspect, the estimating unit weights the second pixel located closer to the center of the product image than the first pixel included in the product image to a weight greater than the first pixel. The number of pixels for each color name candidate is calculated.
　この側面によれば、商品画像において商品の色としてより適切な色が配置されている可能性が高い中央部に近いピクセルにより大きい重み付けがされて色名候補ごとのピクセル数の算出が行われるので、より精度の高い色の推定が可能となる。 According to this aspect, since the pixel near the center, which is likely to have a more appropriate color as the product color in the product image, is more heavily weighted, the number of pixels for each color name candidate is calculated. Therefore, it is possible to estimate the color with higher accuracy.
　さらに別の側面に係る色推定装置は、推定手段は、色名候補に対して設定されたピクセル値に基づき、色名候補ごとの色範囲を重複が生じないように設定する。 In the color estimation apparatus according to still another aspect, the estimation unit sets the color range for each color name candidate so as not to overlap based on the pixel value set for the color name candidate.
　この側面によれば、色名候補に対して設定される色範囲が重複なく設定されるので、抽出された複数の色名候補が近いピクセル値を有する場合であっても、適切に色名候補ごとのピクセル数の算出をすることが可能となる。 According to this aspect, since the color ranges set for the color name candidates are set without duplication, even if the extracted color name candidates have close pixel values, the color name candidates are appropriately displayed. It is possible to calculate the number of pixels for each.
　さらに別の側面に係る色推定装置では、推定手段は、商品画像のピクセルをピクセル値によりクラスタリングし、グループの中央値が色名候補の色範囲に含まれる場合に、該グループに含まれるピクセルの数を該色名候補の色範囲に含まれるピクセル数として、色名候補ごとのピクセル数の算出を行う。 In the color estimation device according to another aspect, the estimation unit clusters the pixels of the product image based on the pixel values, and when the median value of the group is included in the color range of the color name candidate, The number of pixels for each color name candidate is calculated by using the number as the number of pixels included in the color range of the color name candidate.
　商品画像において、商品の色とされる色を有する領域は一定程度の範囲を占めるので、この側面によれば、ピクセル値によりクラスタリングされたピクセルのグループごとに色名候補が対応付けられることにより、同様の色の領域ごとに色名候補が対応付けられ、ピクセル数の算出が行われることとなる。これにより、効率的に精度の高い色の推定を行うことが可能となる。 In the product image, the region having the color that is the color of the product occupies a certain range, so according to this aspect, the color name candidates are associated with each group of pixels clustered by pixel values, Color name candidates are associated with each similar color region, and the number of pixels is calculated. As a result, it is possible to efficiently estimate the color with high accuracy.
　さらに別の側面に係る色推定装置は、推定手段は、算出されたピクセル数が最も多い色名候補、及び該色名候補と算出されたピクセル数の差が所定数以下のピクセル数の色名候補を、商品画像に表された複数の商品のそれぞれの色として推定する。 In the color estimation device according to another aspect, the estimation unit includes a color name candidate having the largest number of calculated pixels, and a color name having a pixel number that is equal to or less than a predetermined number of difference between the color name candidate and the calculated pixel number. Candidates are estimated as the colors of the plurality of products represented in the product image.
　この側面によれば、色が異なる複数の商品が一の商品画像に表されている場合であっても、各商品の色の推定が可能となる。 According to this aspect, it is possible to estimate the color of each product even when a plurality of products having different colors are represented in one product image.
　さらに別の側面に係る色推定装置では、推定手段は、一の商品について、複数の商品画像が対応付けられている場合、一の商品画像ごとに色を推定する。 In the color estimation device according to another aspect, the estimation means estimates a color for each product image when a plurality of product images are associated with one product.
　この側面によれば、一の商品について複数の商品画像が対応付けられている場合において、各商品画像に表された商品の色を推定できる。 According to this aspect, when a plurality of product images are associated with one product, the color of the product represented in each product image can be estimated.
　さらに別の側面に係る色推定装置では、推定手段は、一の商品について、複数の商品画像が対応付けられている場合、複数の色名候補のうちの一の色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出し、最もピクセル数が多い商品画像の色が一の色名候補であると推定する。 In the color estimation device according to another aspect, when a plurality of product images are associated with one product, the estimation unit is set for one color name candidate among the plurality of color name candidates. The number of product image pixels included in the color range that is the range of pixel values is calculated for each of the plurality of product images, and the color of the product image with the largest number of pixels is estimated as one color name candidate. .
　この側面によれば、一の色名候補について、各商品画像の当該色名に相当するピクセル値のピクセルの数が算出され、最もピクセル数が多い商品画像の色として、一の色名候補が推定されるので、抽出された複数の色名候補が近いピクセル値を有する場合であっても、精度良い色の推定が可能となる。 According to this aspect, for one color name candidate, the number of pixels having a pixel value corresponding to the color name of each product image is calculated, and one color name candidate is selected as the color of the product image having the largest number of pixels. Therefore, even when the extracted plurality of color name candidates have similar pixel values, it is possible to accurately estimate the color.
　さらに別の側面に係る色推定装置では、推定手段は、一の商品について、複数の商品画像が対応付けられている場合、色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる一の商品画像のピクセルの数を色名候補ごとに算出し、算出されたピクセル数が最も多い色名候補を、一の商品画像に表された商品の色として推定し、複数の色名候補のうち、いずれの商品画像の商品の色として推定されなかった色名候補が存在する場合に、複数の色名候補のうちの一の色名候補に対して設定されたピクセル値の範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出し、最もピクセル数が多い商品画像の色が一の色名候補であると推定する。 In the color estimation device according to another aspect, the estimation unit, when a plurality of product images are associated with one product, has a color range that is a pixel value range set for the color name candidate. The number of pixels of one product image included is calculated for each color name candidate, and the color name candidate with the largest number of calculated pixels is estimated as the color of the product represented in one product image, and multiple colors The range of pixel values set for one color name candidate among a plurality of color name candidates when there is a color name candidate that is not estimated as the product color of any product image Is calculated for each of the plurality of product images, and the color of the product image having the largest number of pixels is estimated as one color name candidate.
　この側面によれば、一の商品画像について、各色名候補に対して設定された色範囲に含まれるピクセル数が最も多い色名候補を一の商品画像に表された商品の色として推定し、全ての商品画像についての色の推定を実施した結果、商品の色として推定されなかった色が存在する場合に、一の色名候補について、各商品画像の当該色名に相当するピクセル値のピクセルの数が算出され、最もピクセル数が多い商品画像の色名として、一の色名候補が推定される。これにより、一の商品についての複数の商品画像に複数の色名候補が対応付けられる場合において、複数の色名候補が近いピクセル値を有する場合であっても、精度良く商品の色を推定できる。 According to this aspect, for one product image, the color name candidate having the largest number of pixels included in the color range set for each color name candidate is estimated as the color of the product represented in the one product image, As a result of color estimation for all product images, when there is a color that has not been estimated as a product color, for one color name candidate, a pixel having a pixel value corresponding to the color name of each product image The color name candidate is estimated as the color name of the product image with the largest number of pixels. Thereby, when a plurality of color name candidates are associated with a plurality of product images for one product, the color of the product can be accurately estimated even when the plurality of color name candidates have close pixel values. .
　さらに別の側面に係る色推定装置では、抽出手段は、商品情報に商品の色の指定をユーザから受け付けるための欄及び色テキスト情報が含まれる場合、欄の検索により色テキスト情報を色名候補として抽出する。 In the color estimation device according to another aspect, the extraction unit may extract the color text information by searching the field when the product information includes a field for accepting designation of the product color from the user and the color text information. Extract as
　上記側面によれば、適切に色名候補を抽出できる。 According to the above aspect, color name candidates can be extracted appropriately.
　さらに別の側面に係る色推定装置では、出力手段は、商品情報と商品画像とを対応付けて記憶している商品情報記憶手段の商品画像に、推定された色を対応付けて記憶させる。 In the color estimation device according to still another aspect, the output unit stores the estimated color in association with the product image of the product information storage unit that stores the product information and the product image in association with each other.
　上記側面によれば、商品画像に色が対応付けられることとなるので、色に基づく商品画像の処理が実施可能になる。 According to the above aspect, since the color is associated with the product image, the product image can be processed based on the color.
　さらに別の側面に係る色推定装置は、商品の色の指定を含むユーザからの検索要求に応じて、商品情報記憶手段を参照して、指定された色が対応づけられた商品画像を含む検索結果を該ユーザに返信する検索手段をさらに備える。 According to another aspect of the present invention, a color estimation apparatus refers to a product information storage unit in response to a search request from a user including specification of a product color, and includes a product image associated with a specified color. A search means for returning the result to the user is further provided.
　この側面によれば、ユーザからの色を指定した商品画像の検索要求に応じて、その色を有する商品画像をユーザに提供できる。 According to this aspect, the product image having the color can be provided to the user in response to the search request for the product image specifying the color from the user.
　さらに別の側面に係る色推定装置は、商品情報記憶手段において、一の商品について複数の商品画像が対応付けられており、複数の商品画像のうちの一の商品画像が商品を提示するための商品ページにおいて該商品を示すための代表画像として設定されている場合に、検索手段は、代表画像に対応づけられた色と異なる色の指定を含むユーザからの検索要求を受け付けた場合に、指定された色が対応づけられた商品画像を代表画像として含む検索結果を該ユーザに返信する。 In the color estimation device according to another aspect, in the product information storage unit, a plurality of product images are associated with one product, and one product image of the plurality of product images presents the product. When the product page is set as a representative image for indicating the product, the search means specifies when a search request from a user including a color specification different from the color associated with the representative image is received. A search result including the product image associated with the color as a representative image is returned to the user.
　この側面によれば、ユーザからの色の指定を伴う商品の検索要求に対して、検索要求に係る色とは異なる色の画像が当該商品の代表画像として予め設定されている場合であっても、検索要求に係る色の商品画像をユーザに提供できる。 According to this aspect, even when an image having a color different from the color related to the search request is set in advance as a representative image of the product in response to a search request for the product with the color specification from the user The product image of the color related to the search request can be provided to the user.
　本発明の一側面によれば、複数の色が含まれる商品画像に表されている商品の色を精度良く推定することが可能となる。 According to one aspect of the present invention, it is possible to accurately estimate the color of a product represented in a product image including a plurality of colors.
本実施形態に係る色推定装置１の機能的構成を示すブロック図である。It is a block diagram which shows the functional structure of the color estimation apparatus 1 which concerns on this embodiment. 色推定装置のハードウェア構成を示す図である。It is a figure which shows the hardware constitutions of a color estimation apparatus. 商品情報記憶部の構成及び記憶されているデータの例を示す図である。It is a figure which shows the example of a structure of the merchandise information storage part, and the data stored. 商品情報記憶部に記憶された情報に基づき表示される商品ページの例を示す図である。It is a figure which shows the example of the goods page displayed based on the information memorize | stored in the goods information storage part. 商品情報記憶部に記憶された情報に基づき表示される商品ページの例を示す図である。It is a figure which shows the example of the goods page displayed based on the information memorize | stored in the goods information storage part. 商品情報記憶部に記憶された情報に基づき表示される商品ページの例を示す図である。It is a figure which shows the example of the goods page displayed based on the information memorize | stored in the goods information storage part. 色情報記憶部の構成及び記憶されているデータの例を示す図である。It is a figure which shows the example of a structure of the color information storage part, and the data stored. ＨＳＬ色空間を模式的に示す図である。It is a figure which shows an HSL color space typically. 図９（ａ）は、色名候補「レッド」及び「ブルー」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。図９（ｂ）は、色名候補「ブルー」、「グリーン」及び「ホワイト」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。FIG. 9A is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “red” and “blue”. FIG. 9B is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “blue”, “green”, and “white”. 推定された色名が記憶された商品情報記憶部の例を示す図である。It is a figure which shows the example of the goods information storage part in which the estimated color name was memorize | stored. 色空間における色名候補「レッド」、「ダークグリーン」及び「イエローグリーン」の色範囲を２次元で模式的に示す図である。FIG. 4 is a diagram schematically illustrating two-dimensional color ranges of color name candidates “red”, “dark green”, and “yellow green” in a color space. 色推定装置における色推定方法の処理内容の例を示すフローチャートである。It is a flowchart which shows the example of the processing content of the color estimation method in a color estimation apparatus. 色推定プログラムの構成を示す図である。It is a figure which shows the structure of a color estimation program. 商品情報記憶部に記憶されたデータの例を示す図である。It is a figure which shows the example of the data memorize | stored in the merchandise information storage part. 色名候補「ダークブラウン」、「ダークブルー」、「ブラック」にそれぞれについて算出された、各商品画像のそれぞれのピクセルが色範囲に含まれるピクセル数の例を示す図である。It is a figure which shows the example of the pixel number which each pixel of each product image was calculated about color name candidate "dark brown", "dark blue", and "black", respectively, and is contained in a color range. 商品情報記憶部に記憶されたデータの例を示す図である。It is a figure which shows the example of the data memorize | stored in the merchandise information storage part. 図１７（ａ）は、商品画像Ｐ５１について、色名候補「ブラック」及び「ダークブルー」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。図１７（ｂ）は、商品画像Ｐ５２について、色名候補「ブラック」及び「ダークブルー」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。図１７（ｃ）は、色名候補「ブラック」、「ダークブルー」のそれぞれについて算出された、商品画像Ｐ５１，Ｐ５２のそれぞれのピクセルが色範囲に含まれるピクセル数の例を示す図である。FIG. 17A is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P51. FIG. 17B is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P52. FIG. 17C is a diagram illustrating an example of the number of pixels that are calculated for each of the color name candidates “black” and “dark blue” and each pixel of the product images P51 and P52 is included in the color range. 推定部による色名の推定処理のさらに別の例の内容を示すフローチャートである。It is a flowchart which shows the content of the further another example of the estimation process of the color name by an estimation part.
　以下、添付図面を参照しながら本発明の実施形態を詳細に説明する。なお、図面の説明において同一又は同等の要素には同一の符号を付し、重複する説明を省略する。 Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or equivalent elements are denoted by the same reference numerals, and redundant description is omitted.
　図１は、本実施形態に係る色推定装置１の機能的構成を示すブロック図である。本実施形態の色推定装置１は、商品画像に表された商品の色名を推定する装置である。色推定装置１は、例えば、サーバに構成されるが、電子商取引サイトを構成する電子商取引サーバ３と一体に構成されることとしてもよい。電子商取引サーバ３は、インターネットを介した商取引を制御するサーバであって、例えば、ユーザからの商品の検索要求に応じて、検索結果を返信する。 FIG. 1 is a block diagram showing a functional configuration of the color estimation apparatus 1 according to the present embodiment. The color estimation device 1 of the present embodiment is a device that estimates the color name of a product represented in a product image. The color estimation device 1 is configured as a server, for example, but may be configured integrally with the electronic commerce server 3 configuring the electronic commerce site. The electronic commerce server 3 is a server that controls commerce via the Internet, and returns a search result in response to, for example, a search request for a product from a user.
　本実施形態の色推定装置１は、図１に示すように、機能的には、抽出部１１（抽出手段）、推定部１２（推定手段）及び出力部１３（出力手段）を備える。また、色推定装置１の各機能部は、商品情報記憶部２１（商品情報記憶手段）及び色情報記憶部２２といった記憶手段にアクセス可能である。 As shown in FIG. 1, the color estimation apparatus 1 according to the present embodiment functionally includes an extraction unit 11 (extraction unit), an estimation unit 12 (estimation unit), and an output unit 13 (output unit). Each functional unit of the color estimation apparatus 1 can access storage means such as a product information storage unit 21 (product information storage unit) and a color information storage unit 22.
　図２は、色推定装置１のハードウェア構成図である。色推定装置１は、物理的には、図２に示すように、プロセッサにより構成されるＣＰＵ１０１、ＲＡＭ及びＲＯＭといったメモリにより構成される主記憶装置１０２、ハードディスク等で構成される補助記憶装置１０３、ネットワークカード等で構成される通信制御装置１０４、入力デバイスであるキーボード、マウス等の入力装置１０５、ディスプレイ等の出力装置１０６などを含むコンピュータシステムとして構成されている。 FIG. 2 is a hardware configuration diagram of the color estimation apparatus 1. As shown in FIG. 2, the color estimation apparatus 1 physically includes a CPU 101 configured by a processor, a main storage device 102 configured by a memory such as a RAM and a ROM, an auxiliary storage device 103 configured by a hard disk, and the like. The computer system includes a communication control device 104 constituted by a network card or the like, an input device 105 such as a keyboard and mouse as input devices, an output device 106 such as a display, and the like.
　図１に示した各機能は、図２に示すＣＰＵ１０１、主記憶装置１０２等のハードウェア上に所定のコンピュータソフトウェア（色推定プログラム）を読み込ませることにより、ＣＰＵ１０１の制御のもとで通信制御装置１０４、入力装置１０５、出力装置１０６を動作させるとともに、主記憶装置１０２や補助記憶装置１０３におけるデータの読み出し及び書き込みを行うことで実現される。処理に必要なデータやデータベースは主記憶装置１０２や補助記憶装置１０３内に格納される。 Each function shown in FIG. 1 is a communication control device under the control of the CPU 101 by reading predetermined computer software (color estimation program) on the hardware such as the CPU 101 and the main storage device 102 shown in FIG. 104, the input device 105, and the output device 106 are operated, and data is read and written in the main storage device 102 and the auxiliary storage device 103. Data and databases necessary for processing are stored in the main storage device 102 and the auxiliary storage device 103.
　続いて、色推定装置１の機能部を説明する。抽出部１１は、商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を商品の色名候補として抽出する部分である。商品情報は、ユーザに提示するための当該商品に関するテキスト情報を含む。商品に関するテキスト情報は、例えば、商品に関する説明文、商品ページにおいて商品の色の指定を受け付けるための欄を伴う色名等である。 Subsequently, the functional unit of the color estimation apparatus 1 will be described. The extraction unit 11 is a part that extracts a plurality of color text information representing color names as product color name candidates from the product information associated with the product image representing the product. The product information includes text information regarding the product to be presented to the user. The text information related to the product is, for example, an explanatory note related to the product, a color name accompanied with a column for accepting designation of the color of the product on the product page.
　本実施形態では、商品情報記憶部２１は、電子商取引に供される商品に関する情報を記憶しており、商品を識別する商品ＩＤに、商品情報及び商品画像を対応付けて記憶している。図３は、商品情報記憶部２１の構成及び記憶されているデータの例を示す図である。図３に示すように、例えば、商品情報記憶部２１は、商品ＩＤ（Ｍ１）に対応付けて、商品情報Ｔ１及び商品画像Ｐ１を記憶している。商品情報Ｔ１は、その内容として、「色は、レッドとブルーがあります。」というユーザに提示するためのテキスト情報を含む。 In the present embodiment, the product information storage unit 21 stores information related to products used for electronic commerce, and stores product information and product images in association with product IDs for identifying products. FIG. 3 is a diagram illustrating a configuration of the product information storage unit 21 and an example of stored data. As illustrated in FIG. 3, for example, the product information storage unit 21 stores product information T1 and a product image P1 in association with the product ID (M1). The product information T1 includes, as its content, text information to be presented to the user that “there are red and blue colors”.
　また、商品情報記憶部２１は、商品ＩＤ（Ｍ２）に対応付けて、商品情報Ｔ２及び商品画像Ｐ２を記憶している。商品情報Ｔ２は、その内容として、「・・・ブルー、グリーン、ホワイトの三色があり、・・」というテキスト情報を含む。 The product information storage unit 21 stores product information T2 and a product image P2 in association with the product ID (M2). The product information T2 includes text information “... There are three colors of blue, green and white,...
　また、商品情報記憶部２１は、商品ＩＤ（Ｍ３）に対応付けて、商品情報Ｔ３及び複数の商品画像Ｐ３１，Ｐ３２，Ｐ３３を記憶している。商品情報Ｔ３は、その内容として、商品ページにおいて商品の色の指定を受け付けるための欄を伴う色テキスト情報「ブルー」、「レッド」及び「イエロー」を含む。これらの欄及び色テキスト情報は、ユーザに提示するためのものである。また、商品画像Ｐ３１，Ｐ３２，Ｐ３３のうちの商品画像Ｐ３１が、商品ＩＤ（Ｍ３）の商品を商品ページとして表示する際の代表画像に設定されている。 The product information storage unit 21 stores product information T3 and a plurality of product images P31, P32, and P33 in association with the product ID (M3). The product information T3 includes, as its contents, color text information “blue”, “red”, and “yellow” with a column for accepting designation of the product color on the product page. These fields and color text information are for presentation to the user. The product image P31 among the product images P31, P32, and P33 is set as a representative image when the product with the product ID (M3) is displayed as a product page.
　図４は、商品情報記憶部２１に記憶された情報に基づき表示される商品ＩＤ（Ｍ１）の商品ページの例を示す図である。図４に示すように、商品ＩＤ（Ｍ１）の商品ページは、商品を表す商品画像Ｐ１及び商品に関する説明文「色は、レッドとブルーがあります。」からなる商品情報Ｔ１を含む。即ち、商品情報Ｔ１は、色名を表す色テキスト情報「レッド」及び「ブルー」を含む。これらの色テキスト情報は、商品画像Ｐ１に表された商品の色に該当する可能性が高い。 FIG. 4 is a diagram illustrating an example of the product page of the product ID (M1) displayed based on the information stored in the product information storage unit 21. As shown in FIG. 4, the product page with the product ID (M1) includes product information T1 including a product image P1 representing the product and an explanatory text “There are red and blue colors”. That is, the product information T1 includes color text information “red” and “blue” representing the color name. There is a high possibility that the color text information corresponds to the color of the product shown in the product image P1.
　図５は、商品情報記憶部２１に記憶された情報に基づき表示される商品ＩＤ（Ｍ２）の商品ページの例を示す図である。図５に示すように、商品ＩＤ（Ｍ２）の商品ページは、商品を表す商品画像Ｐ２及び商品に関する説明文「・・・ブルー、グリーン、ホワイトの三色があり、・・・。」からなる商品情報Ｔ２を含む。即ち、商品情報Ｔ２は、色名を表す色テキスト情報「ブルー」、「グリーン」及び「ホワイト」を含む。これらの色テキスト情報は、商品画像Ｐ２に表された商品の色に該当する可能性が高い。 FIG. 5 is a diagram illustrating an example of the product page of the product ID (M2) displayed based on the information stored in the product information storage unit 21. As shown in FIG. 5, the product page of the product ID (M2) is composed of a product image P2 representing the product and a description regarding the product “... There are three colors of blue, green, and white,. Product information T2 is included. That is, the product information T2 includes color text information “blue”, “green”, and “white” representing the color name. There is a high possibility that the color text information corresponds to the color of the product shown in the product image P2.
　図６は、商品情報記憶部２１に記憶された情報に基づき表示される商品ＩＤ（Ｍ３）の商品ページの例を示す図である。図６に示すように、商品ＩＤ（Ｍ３）の商品ページは、代表画像として設定された商品画像Ｐ３１を含む。また、商品情報記憶部２１において記憶されている商品画像のうち、代表画像以外の商品画像Ｐ３２，Ｐ３３は、サムネイル形式で表示されることとしてもよい。また、この商品ページは、商品の色の指定を受け付けるための複数の欄及び各欄に対応する色テキスト情報「ブルー」、「レッド」及び「イエロー」を含む商品情報Ｔ３を含む。即ち、商品画像Ｐ３１に表された商品の色は、「ブルー」、「レッド」及び「イエロー」のいずれかである可能性が高い。 FIG. 6 is a diagram illustrating an example of the product page of the product ID (M3) displayed based on the information stored in the product information storage unit 21. As shown in FIG. 6, the product page with the product ID (M3) includes a product image P31 set as a representative image. Of the product images stored in the product information storage unit 21, the product images P32 and P33 other than the representative image may be displayed in a thumbnail format. In addition, the product page includes a plurality of columns for receiving designation of product colors and product information T3 including color text information “blue”, “red”, and “yellow” corresponding to each column. That is, there is a high possibility that the color of the product shown in the product image P31 is any one of “blue”, “red”, and “yellow”.
　抽出部１１は、図３並びに図４、図５及び図６に示した商品情報から、色テキスト情報を当該商品の色名候補として抽出する。抽出部１１は、例えば、後に説明する色情報記憶部２２の色名を辞書として参照し、色テキスト情報を抽出できる。具体的には、抽出部１１は、商品情報Ｔ１から、色テキスト情報「レッド」，「ブルー」を商品ＩＤ（Ｍ１）の商品の色名候補として抽出する。また、抽出部１１は、商品情報Ｔ２から、商品の色の指定を受け付けるための複数の欄から、色テキスト情報「ブルー」，「グリーン」，「ホワイト」を商品ＩＤ（Ｍ２）の商品の色名候補として抽出する。また、抽出部１１は、商品情報Ｔ３から、色テキスト情報「ブルー」，「レッド」，「イエロー」を商品ＩＤ（Ｍ３）の商品の色名候補として抽出する。 The extraction unit 11 extracts color text information as color name candidates for the product from the product information shown in FIGS. 3, 4, 5, and 6. For example, the extraction unit 11 can extract color text information by referring to a color name in the color information storage unit 22 described later as a dictionary. Specifically, the extraction unit 11 extracts the color text information “red” and “blue” from the product information T1 as color name candidates for the product with the product ID (M1). Further, the extraction unit 11 uses the color text information “blue”, “green”, and “white” from the plurality of fields for receiving the designation of the product color from the product information T2, and the product color of the product ID (M2). Extract as name candidates. Further, the extraction unit 11 extracts the color text information “blue”, “red”, and “yellow” from the product information T3 as product color name candidates of the product ID (M3).
　なお、商品情報から抽出された色テキスト情報には、同じ色を示す色名であっても、例えば「黒」、「ブラック」のように、ばらつきが含まれる場合がある。このような場合に、抽出部１１は、抽出した色テキスト情報を規格化して、色名候補とすることとしてもよい。即ち、商品情報から「黒」及び「ブラック」という色テキスト情報が抽出された場合に、抽出部１１は、これらの色テキスト情報を、「ブラック」に規格化して色名候補とする。 Note that the color text information extracted from the product information may include variations such as “black” and “black” even if the color names indicate the same color. In such a case, the extraction unit 11 may normalize the extracted color text information and set it as a color name candidate. That is, when the color text information “black” and “black” is extracted from the product information, the extraction unit 11 normalizes these color text information to “black” and sets them as color name candidates.
　推定部１２は、抽出部１１により抽出された色名候補に対して設定されたピクセル値と、商品画像に含まれるピクセルのピクセル値に基づき、商品画像に表された商品の色を推定する部分である。 The estimation unit 12 is a unit that estimates the color of the product represented in the product image based on the pixel value set for the color name candidate extracted by the extraction unit 11 and the pixel value of the pixel included in the product image. It is.
　本実施形態では、推定部１２は、抽出部１１により抽出された色名候補の中から商品画像に表された商品の色名を推定する。即ち、推定部１２は、色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる商品画像のピクセルの数を色名候補ごとに算出し、算出されたピクセル数が最も多い色名候補を、商品画像に表された商品の色名として推定する。 In this embodiment, the estimation unit 12 estimates the color name of the product represented in the product image from the color name candidates extracted by the extraction unit 11. That is, the estimating unit 12 calculates the number of pixels of the product image included in the color range that is the range of pixel values set for the color name candidate for each color name candidate, and the calculated number of pixels is the largest. The color name candidate is estimated as the color name of the product shown in the product image.
　本実施形態では、推定部１２は、色情報記憶部２２を参照して、色名候補に対応する色範囲を取得する。図７は、色情報記憶部２２の構成及び記憶されているデータの例を示す図である。色情報記憶部２２は、色名ごとに設定されたピクセル値を記憶している。具体的には、色情報記憶部２２は、色名ごとに設定された色範囲を記憶している。また、色情報記憶部２２は、色名ごとに中央値をさらに記憶していてもよい。なお、本実施形態では、ピクセル値は、ＨＳＬ色空間で表されるが、その他のパラメータ（例えば、ＨＳＶ空間、ＲＧＢ空間等）で表されることを排除しない。例えば、色名「レッド」の色範囲は、（Ｈ１Ｌ～Ｈ１Ｈ，Ｓ１Ｌ～Ｓ１Ｈ，Ｌ１Ｌ～Ｌ１Ｈ）であり、中央値は、（Ｈ１Ｍ，Ｓ１Ｍ，Ｌ１Ｍ）である。 In the present embodiment, the estimation unit 12 refers to the color information storage unit 22 and acquires a color range corresponding to a color name candidate. FIG. 7 is a diagram illustrating a configuration of the color information storage unit 22 and an example of stored data. The color information storage unit 22 stores pixel values set for each color name. Specifically, the color information storage unit 22 stores a color range set for each color name. Further, the color information storage unit 22 may further store a median value for each color name. In the present embodiment, the pixel value is expressed in the HSL color space, but it is not excluded that it is expressed in other parameters (for example, HSV space, RGB space, etc.). For example, the color range of the color name “red” is (H 1L to H 1H , S 1L to S 1H , L 1L to L 1H ), and the median is (H 1M , S 1M , L 1M ). .
　図８を参照して、推定部１２による、商品ＩＤ（Ｍ１）の商品画像の色名の推定処理を具体的に説明する。図８は、ＨＳＬ色空間を模式的に示す図である。推定部１２は、商品情報Ｔ１から抽出された色名候補「レッド」及び「ブルー」の色範囲を色情報記憶部２２から取得する。図８において、色名「レッド」の色範囲は、色範囲ＣＲにより示される。また、色名「ブルー」の色範囲は、色範囲ＣＢにより示される。推定部１２は、色名候補「レッド」の色範囲ＣＲ及び色名候補「ブルー」の色範囲ＣＢのそれぞれに含まれる商品画像Ｐ１のピクセルの数を算出する。図９（ａ）は、色名候補「レッド」及び「ブルー」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。図９（ａ）に示すように、商品画像Ｐ１のピクセルのうち、色名候補「レッド」の色範囲ＣＲにピクセル値が含まれるピクセル数は１５０であり、色名候補「ブルー」の色範囲ＣＢにピクセル値が含まれるピクセル数は８５である。従って、推定部１２は、商品画像Ｐ１に表された商品の色名を「レッド」と推定する。このように、推定部１２は、商品の色を表している可能性が最も高い色名候補を商品の色名として推定できる。 With reference to FIG. 8, the estimation process of the color name of the product image of the product ID (M1) by the estimation unit 12 will be specifically described. FIG. 8 is a diagram schematically showing the HSL color space. The estimation unit 12 acquires the color name candidates “red” and “blue” extracted from the product information T1 from the color information storage unit 22. In FIG. 8, the color range of the color name “red” is indicated by the color range CR. The color range of the color name “blue” is indicated by the color range CB. The estimation unit 12 calculates the number of pixels of the product image P1 included in each of the color range CR of the color name candidate “red” and the color range CB of the color name candidate “blue”. FIG. 9A is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “red” and “blue”. As shown in FIG. 9A, among the pixels of the product image P1, the number of pixels whose pixel values are included in the color range CR of the color name candidate “red” is 150, and the color range of the color name candidate “blue” The number of pixels including a pixel value in CB is 85. Accordingly, the estimation unit 12 estimates the color name of the product represented in the product image P1 as “red”. In this way, the estimation unit 12 can estimate the color name candidate that is most likely to represent the color of the product as the color name of the product.
　なお、ピクセル数の算出に際して、推定部１２は、商品画像に含まれる第１のピクセルより商品画像の中央部に近い位置にある第２のピクセルに対して、第１のピクセルより大きい重み付けをして、色名候補ごとのピクセル数の算出を行うこととしてもよい。具体的には、推定部１２は、商品画像の中央部を含む所定の範囲内に位置するピクセルに重み付けをしてピクセル数の算出をしてもよいし、中心部からの距離に比例して重み付けをしてピクセル数の算出をしてもよい。これにより、商品画像において商品の色名としてより適切な色が配置されている可能性が高い中央部に近いピクセルにより大きい重み付けがされて色名候補ごとのピクセル数の算出が行われるので、より精度の高い色名の推定が可能となる。 When calculating the number of pixels, the estimation unit 12 weights the second pixel closer to the center of the product image than the first pixel included in the product image, more than the first pixel. Thus, the number of pixels for each color name candidate may be calculated. Specifically, the estimation unit 12 may calculate the number of pixels by weighting pixels located within a predetermined range including the central part of the product image, or in proportion to the distance from the central part. The number of pixels may be calculated by weighting. As a result, more weight is given to the pixels near the central portion where there is a high possibility that a more appropriate color is arranged as the color name of the product in the product image, and the number of pixels for each color name candidate is calculated. It is possible to estimate a color name with high accuracy.
　また、推定部１２は、商品画像における商品が表された領域以外の領域のピクセルを、ピクセル数の算出の対象から除外してもよい。商品画像における商品が表された領域以外の領域は、周知の画像処理技術により抽出可能である。例えば、推定部１２は、背景領域のピクセルを、ピクセル数の算出の対象から除外してもよい。 Further, the estimation unit 12 may exclude pixels in a region other than the region in which the product is represented in the product image from the target for calculating the number of pixels. A region other than the region where the product is represented in the product image can be extracted by a known image processing technique. For example, the estimation unit 12 may exclude the pixels in the background area from the target for calculating the number of pixels.
　商品ＩＤ（Ｍ２）の商品について（図３及び図５参照）、推定部１２は、商品情報Ｔ２から抽出された色名候補「ブルー」、「グリーン」及び「ホワイト」の色範囲を色情報記憶部２２から取得する。そして、推定部１２は、色名候補「ブルー」、「グリーン」及び「ホワイト」のそれぞれの色範囲に含まれる商品画像Ｐ２のピクセルの数を算出する。図９（ｂ）は、色名候補「ブルー」、「グリーン」及び「ホワイト」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。図９（ｂ）に示すように、商品画像Ｐ２のピクセルのうち、色名候補「ブルー」の色範囲にピクセル値が含まれるピクセル数は１８０であり、色名候補「グリーン」の色範囲にピクセル値が含まれるピクセル数は１７５であり、色名候補「ホワイト」の色範囲にピクセル値が含まれるピクセル数は４５である。 For the product with the product ID (M2) (see FIGS. 3 and 5), the estimation unit 12 stores the color range of the color name candidates “blue”, “green”, and “white” extracted from the product information T2 as color information. Obtained from the unit 22. Then, the estimation unit 12 calculates the number of pixels of the product image P2 included in the color ranges of the color name candidates “blue”, “green”, and “white”. FIG. 9B is a diagram illustrating a calculation result of the number of pixels included in each color range of the color name candidates “blue”, “green”, and “white”. As shown in FIG. 9B, among the pixels of the product image P2, the number of pixels whose pixel value is included in the color range of the color name candidate “blue” is 180, and the color range of the color name candidate “green” The number of pixels including the pixel value is 175, and the number of pixels including the pixel value in the color range of the color name candidate “white” is 45.
　ここで、推定部１２は、算出されたピクセル数が最も多い色名候補、及び該色名候補と算出されたピクセル数の差が所定数以下のピクセル数の色名候補を、商品画像に表された複数の商品のそれぞれの色名として推定することができる。複数の色名候補を商品画像に表された複数の商品のそれぞれの色名として推定するためのピクセル数の差に関する所定数が１０であるとして、商品ＩＤ（Ｍ２）の商品の例で具体的に説明する。 Here, the estimation unit 12 displays a color name candidate having the largest number of calculated pixels and a color name candidate having a number of pixels whose difference between the color name candidate and the calculated number of pixels is a predetermined number or less in the product image. It can be estimated as the color name of each of a plurality of products. Specifically, in the example of the product with the product ID (M2), the predetermined number regarding the difference in the number of pixels for estimating the plurality of color name candidates as the color names of the plurality of products represented in the product image is 10. Explained.
　色名候補「ブルー」、「グリーン」及び「ホワイト」のうち、それぞれの色範囲に含まれる商品画像Ｐ２のピクセルのピクセル数が最も多い色名候補は、「ブルー」（ピクセル数１８０）であり、このピクセル数に対して、「グリーン」の色範囲に含まれるピクセル数が１７５であって、その差が１０以下であるので、推定部１２は、「ブルー」及び「グリーン」を商品画像Ｐ２に表された複数の商品のそれぞれの色名として推定する。このような推定処理により、色が異なる複数の商品が一の商品画像に表されている場合であっても、各商品の色名の推定が可能となる。 Among the color name candidates “blue”, “green”, and “white”, the color name candidate having the largest number of pixels of the product image P2 included in each color range is “blue” (180 pixels). Since the number of pixels included in the color range of “green” is 175 with respect to this number of pixels and the difference is 10 or less, the estimation unit 12 assigns “blue” and “green” to the product image P2. Estimated as the color name of each of the multiple products represented in. By such an estimation process, it is possible to estimate the color name of each product even when a plurality of products with different colors are represented in one product image.
　商品ＩＤ（Ｍ３）の商品について（図３及び図６参照）、推定部１２は、商品情報Ｔ３から抽出された色名候補「ブルー」、「レッド」及び「イエロー」の色範囲を色情報記憶部２２から取得する。ここで、推定部１２は、一の商品について、複数の商品画像が対応付けられている場合、一の商品画像ごとに色名を推定する。具体的には、商品ＩＤ（Ｍ３）の商品について、商品情報Ｔ３と複数の商品画像Ｐ３１，Ｐ３２，Ｐ３３が対応付けられているので、推定部１２は、商品画像Ｐ３１，Ｐ３２，Ｐ３３のそれぞれの商品画像ごとに、商品画像Ｐ１における色名の推定と同様に、色名候補「ブルー」、「レッド」及び「イエロー」のそれぞれの色範囲に含まれるピクセルの数を算出して、各商品画像Ｐ３１，Ｐ３２，Ｐ３３に表された商品の色名を推定する。具体的なピクセル数を示した推定処理の説明は省略する。 For the product with the product ID (M3) (see FIGS. 3 and 6), the estimation unit 12 stores the color range of the color name candidates “blue”, “red”, and “yellow” extracted from the product information T3 as color information. Obtained from the unit 22. Here, when a plurality of product images are associated with one product, the estimation unit 12 estimates a color name for each product image. Specifically, because the product information T3 and the plurality of product images P31, P32, and P33 are associated with the product with the product ID (M3), the estimation unit 12 determines each of the product images P31, P32, and P33. For each product image, as in the estimation of the color name in the product image P1, the number of pixels included in each color range of the color name candidates “blue”, “red”, and “yellow” is calculated, and each product image The color name of the product represented by P31, P32, and P33 is estimated. The description of the estimation process indicating the specific number of pixels is omitted.
　なお、以上説明した色名の推定では、推定部１２は、色範囲に含まれる商品画像のピクセルの数を色名候補ごとに算出することとしたが、ピクセルの数に代えて、商品画像のピクセルに占める色範囲に含まれるピクセルの割合を算出して商品画像に表された商品の色名として推定することとしてもよい。 In the color name estimation described above, the estimation unit 12 calculates the number of pixels of the product image included in the color range for each color name candidate, but instead of the number of pixels, the product image The ratio of the pixels included in the color range occupied by the pixels may be calculated and estimated as the color name of the product displayed in the product image.
　なお、本実施形態では、抽出部１１により抽出された色名候補のピクセル値に基づいて、推定部１２が商品画像の商品の色名を推定することとしたが、推定部１２は、ピクセル値として特定される色を推定することとしてもよい。つまり、推定する色として、テキスト情報である色名を推定してもよいし、表示される色を示すピクセル値を推定してもよい。 In the present embodiment, the estimation unit 12 estimates the color name of the product image of the product image based on the pixel value of the color name candidate extracted by the extraction unit 11. It is good also as estimating the color specified as. That is, as the color to be estimated, a color name that is text information may be estimated, or a pixel value indicating a displayed color may be estimated.
　再び図１を参照して、出力部１３は、推定部１２により推定された色名を出力する部分である。具体的には、出力部１３は、商品情報と商品画像とを対応付けて記憶している商品情報記憶部２１の商品画像に、推定された色名を対応付けて記憶させる。図１０は、推定された色名が記憶された商品情報記憶部２１の例を示す図である。図１０に示されるように、出力部１３は、商品画像Ｐ１に推定された色名「レッド」を対応付けて記憶させる。また、出力部１３は、商品画像Ｐ２に推定された色名「ブルー」及び「グリーン」を対応付けて記憶させる。また、出力部１３は、商品画像Ｐ３１，Ｐ３２，Ｐ３３のそれぞれに、推定された色名「ブルー」、「レッド」及び「イエロー」を対応付けて記憶させる。このように、商品画像に色名が対応付けられることとなるので、色名に基づく商品画像の処理が実施可能になる。例えば、色名の指定による商品画像の抽出が可能となる。 Referring to FIG. 1 again, the output unit 13 is a part that outputs the color name estimated by the estimation unit 12. Specifically, the output unit 13 stores the estimated color name in association with the product image in the product information storage unit 21 that stores the product information and the product image in association with each other. FIG. 10 is a diagram illustrating an example of the product information storage unit 21 in which the estimated color names are stored. As illustrated in FIG. 10, the output unit 13 stores the estimated color name “red” in association with the product image P1. Further, the output unit 13 stores the estimated color names “blue” and “green” in association with the product image P2. Further, the output unit 13 stores the estimated color names “blue”, “red”, and “yellow” in association with the product images P31, P32, and P33, respectively. As described above, since the color name is associated with the product image, the product image based on the color name can be processed. For example, a product image can be extracted by specifying a color name.
　次に、推定部１２による処理のいくつかのバリエーションを説明する。例えば、抽出部１１により、色名候補「レッド」、「ダークグリーン」及び「イエローグリーン」が抽出された場合、推定部１２は、色名候補「レッド」、「ダークグリーン」及び「イエローグリーン」の色範囲を色情報記憶部２２から取得する。図１１（ａ）は、色空間における色名候補「レッド」、「ダークグリーン」及び「イエローグリーン」の色範囲を２次元で模式的に示す図である。図１１（ａ）に示すように、レッドの色範囲ＣＲと、ダークグリーンの色範囲ＣＤＧ１及びイエローグリーンの色範囲ＣＹＧ１とは互いに離れているが、「ダークグリーン」及び「イエローグリーン」を示すピクセル値は互いに近い値を有するので、ダークグリーンの色範囲ＣＤＧ１とイエローグリーンの色範囲ＣＹＧ１とは重複している。 Next, some variations of processing by the estimation unit 12 will be described. For example, when the extraction unit 11 extracts color name candidates “red”, “dark green”, and “yellow green”, the estimation unit 12 determines the color name candidates “red”, “dark green”, and “yellow green”. Are acquired from the color information storage unit 22. FIG. 11A is a diagram schematically illustrating two-dimensional color ranges of candidate color names “red”, “dark green”, and “yellow green” in the color space. As shown in FIG. 11A, the red color range CR and the dark green color range CDG 1 and the yellow green color range CYG 1 are separated from each other, but “dark green” and “yellow green” are Since the pixel values shown are close to each other, the dark green color range CDG 1 and the yellow green color range CYG 1 overlap.
　このような場合に、推定部１２は、色名候補に対して設定されたピクセル値に基づき、色名候補ごとの色範囲を重複が生じないように設定することができる。具体的には、例えば、推定部１２は、色情報記憶部２２から、「ダークグリーン」及び「イエローグリーン」の色範囲の中央値を取得し、中央値間の距離の２分の１を色範囲の半径とする。図１１（ｂ）は、重複が生じないように設定された「レッド」、「ダークグリーン」及び「イエローグリーン」の色範囲を２次元で模式的に示す図である。図１１（ｂ）に示すように、ダークグリーンの色範囲ＣＤＧ２とイエローグリーンの色範囲ＣＹＧ２との重複が解消されている。このように色範囲を設定することにより、色名候補に対して設定される色範囲が重複なく設定されるので、抽出された複数の色名候補が近いピクセル値を有する場合であっても、適切に色名候補ごとのピクセル数の算出をすることが可能となる。 In such a case, the estimation unit 12 can set the color range for each color name candidate so as not to overlap based on the pixel value set for the color name candidate. Specifically, for example, the estimation unit 12 acquires the median value of the color range of “dark green” and “yellow green” from the color information storage unit 22, and uses half the distance between the median values as the color. The radius of the range. FIG. 11B is a diagram schematically illustrating two-dimensionally the color ranges of “red”, “dark green”, and “yellow green” set so as not to overlap. As shown in FIG. 11B, the overlap between the dark green color range CDG 2 and the yellow green color range CYG 2 is eliminated. By setting the color range in this way, the color ranges set for the color name candidates are set without duplication, so even if the extracted color name candidates have close pixel values, It is possible to appropriately calculate the number of pixels for each color name candidate.
　次に、推定部１２によるピクセル数の算出処理のバリエーションを説明する。商品画像において、商品の色名とされる色を有するピクセルの領域は一定程度の範囲を占める。このような事情に鑑みて、推定部１２は、商品画像のピクセルをピクセル値によりクラスタリングし、クラスタリングされたグループの中央値が色名候補の色範囲に含まれる場合に、そのグループに含まれるピクセルの数をその色名候補の色範囲に含まれるピクセル数として、色名候補ごとのピクセル数の算出を行うこととしてもよい。このようなピクセル数の算出では、ピクセル値によりクラスタリングされたピクセルのグループごとに色名候補が対応付けられることにより、同様の色の領域ごとに色名候補が対応付けられ、ピクセル数の算出が行われることとなる。これにより、効率的に精度の高い色名の推定を行うことが可能となる。 Next, variations of the pixel count calculation process by the estimation unit 12 will be described. In the product image, a pixel region having a color which is a color name of the product occupies a certain range. In view of such circumstances, the estimation unit 12 clusters the pixels of the product image based on the pixel values, and when the median value of the clustered group is included in the color range of the color name candidate, the pixels included in the group The number of pixels for each color name candidate may be calculated using the number of pixels as the number of pixels included in the color range of the color name candidate. In such calculation of the number of pixels, color name candidates are associated with each group of pixels clustered according to pixel values, so that color name candidates are associated with each similar color region, and the number of pixels is calculated. Will be done. This makes it possible to estimate a color name with high accuracy and efficiency.
　上述のとおり、色推定装置１は、電子商取引サイトを構成する電子商取引サーバ３と一体に構成されることとしてもよい（図１参照）。そのような場合において、色推定装置１は、商品の色名の指定を含むユーザからの検索要求に応じて、商品情報記憶部２１を参照して、指定された色名が対応づけられた商品画像を含む検索結果をユーザに返信する検索部３１をさらに備えることとしてもよい。 As described above, the color estimation device 1 may be configured integrally with the electronic commerce server 3 constituting the electronic commerce site (see FIG. 1). In such a case, the color estimation apparatus 1 refers to the product information storage unit 21 in response to a search request from the user including specification of the color name of the product, and the product associated with the specified color name. It is good also as providing the search part 31 which returns the search result containing an image to a user.
　具体的には、ユーザの端末装置（図示せず）からネットワークを介して、商品の色名（例えば「グリーン」）の指定を含む商品の検索要求を受信した場合に、検索部３１は、図１０に示した商品情報記憶部２１を参照して、色名「グリーン」が対応付けられた商品画像Ｐ２を含む商品ＩＤ（Ｍ２）の商品に関する情報を検索結果としてユーザの端末装置に返信する。 Specifically, when a search request for a product including designation of a product color name (for example, “green”) is received from a user terminal device (not shown) via the network, the search unit 31 Referring to the product information storage unit 21 shown in FIG. 10, information related to the product with the product ID (M2) including the product image P2 associated with the color name “green” is returned as a search result to the user's terminal device.
　また、商品ＩＤ（Ｍ３）の商品が該当する「シャツ」というキーワードと共に、商品ＩＤ（Ｍ３）の代表画像に対応づけられた色名「ブルー」と異なる色名「レッド」を含む検索要求を受信した場合には（図１０参照）、検索部３１は、指定された色名「レッド」が対応づけられた商品画像Ｐ３２を代表画像として含む検索結果をユーザに返信することとしてもよい。即ち、図６を参照して説明すると、検索部３１は、商品画像Ｐ３１に代えて、商品画像Ｐ３２を代表画像として表示させる商品ページをユーザの端末装置に返信する。このような処理を行うことにより、ユーザからの色名の指定を伴う商品の検索要求に対して、検索要求に係る色名とは異なる色名の画像が当該商品の代表画像として予め設定されている場合であっても、検索要求に係る色名の商品画像をユーザに提供できる。 In addition, a search request including a keyword “shirt” corresponding to the product with the product ID (M3) and a color name “red” different from the color name “blue” associated with the representative image of the product ID (M3) is received. In such a case (see FIG. 10), the search unit 31 may return a search result including the product image P32 associated with the designated color name “red” as a representative image to the user. That is, with reference to FIG. 6, the search unit 31 returns a product page that displays the product image P32 as a representative image instead of the product image P31 to the user's terminal device. By performing such processing, an image having a color name different from the color name related to the search request is set in advance as a representative image of the product in response to a search request for the product accompanied by designation of the color name from the user. Even if it exists, the product image of the color name which concerns on a search request can be provided to a user.
　次に、図１２を参照して、本実施形態の色推定方法について説明する。図１２は、図１に示した色推定装置１における色推定方法の処理内容の例を示すフローチャートである。 Next, the color estimation method of this embodiment will be described with reference to FIG. FIG. 12 is a flowchart showing an example of processing contents of the color estimation method in the color estimation apparatus 1 shown in FIG.
　まず、抽出部１１は、商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を商品の色名候補として抽出する（Ｓ１）。 First, the extraction unit 11 extracts a plurality of color text information representing color names as product color name candidates from the product information associated with the product image (S1).
　次に、推定部１２は、ステップＳ１において抽出部１１により抽出された色名候補に基づき設定されたピクセル値と、商品画像に含まれるピクセルのピクセル値に基づき、商品画像に表された商品の色名を色名候補の中から推定する。 Next, the estimation unit 12 sets the product value represented in the product image based on the pixel value set based on the color name candidate extracted by the extraction unit 11 in step S1 and the pixel value of the pixel included in the product image. A color name is estimated from among color name candidates.
　具体的には、推定部１２は、色名候補に対して設定された色範囲に含まれる商品画像のピクセルの数を色名候補ごとに算出する（Ｓ２）。そして、推定部１２は、算出されたピクセル数が最も多い色名候補を、商品画像に表された商品の色名として推定する（Ｓ３）。そして、出力部１３は、推定部１２により推定された色名を出力する（Ｓ４）。 Specifically, the estimation unit 12 calculates the number of pixels of the product image included in the color range set for the color name candidate for each color name candidate (S2). Then, the estimation unit 12 estimates the color name candidate having the largest number of calculated pixels as the color name of the product represented in the product image (S3). Then, the output unit 13 outputs the color name estimated by the estimation unit 12 (S4).
　次に、図１３を参照して、コンピュータを色推定装置１として機能させるための色推定プログラムを説明する。色推定プログラム１ｐは、メインモジュールｍ１０、抽出モジュールｍ１１、推定モジュールｍ１２及び出力モジュールｍ１３を備える。また、色推定プログラム１ｐは、検索モジュール（図示せず）を更に備えることとしてもよい。 Next, a color estimation program for causing a computer to function as the color estimation apparatus 1 will be described with reference to FIG. The color estimation program 1p includes a main module m10, an extraction module m11, an estimation module m12, and an output module m13. The color estimation program 1p may further include a search module (not shown).
　メインモジュールｍ１０は、色推定処理を統括的に制御する部分である。抽出モジュールｍ１１、推定モジュールｍ１２及び出力モジュールｍ１３並びに検索モジュールを実行することにより実現される機能はそれぞれ、図１に示される色推定装置１の抽出部１１、推定部１２及び出力部１３並びに検索部３１の機能と同様である。 The main module m10 is a part that comprehensively controls the color estimation process. The functions realized by executing the extraction module m11, the estimation module m12, the output module m13, and the search module are respectively the extraction unit 11, the estimation unit 12, the output unit 13, and the search unit of the color estimation apparatus 1 shown in FIG. This is the same as the function 31.
　色推定プログラム１ｐは、例えば、ＣＤ－ＲＯＭやＤＶＤ－ＲＯＭまたは半導体メモリ等の記憶媒体１ｄによって提供される。また、色推定プログラム１ｐは、搬送波に重畳されたコンピュータデータ信号として通信ネットワークを介して提供されてもよい。 The color estimation program 1p is provided by a storage medium 1d such as a CD-ROM, a DVD-ROM, or a semiconductor memory, for example. The color estimation program 1p may be provided via a communication network as a computer data signal superimposed on a carrier wave.
　以上説明した本実施形態の色推定装置１、色推定方法及び色推定プログラム１ｐによれば、商品画像に表された商品の色を示すテキストが含まれている商品情報から色名候補が抽出され、商品画像に含まれるピクセルのピクセル値及び数に基づき、色名候補の中から商品の色名が推定されるので、誤った色名が商品の色名として推定されることが抑制され、精度の良い色名の推定が実現される。 According to the color estimation device 1, the color estimation method, and the color estimation program 1p of the present embodiment described above, color name candidates are extracted from product information including text indicating the color of the product represented in the product image. Since the color name of the product is estimated from the color name candidates based on the pixel value and the number of pixels included in the product image, it is possible to prevent an incorrect color name from being estimated as the color name of the product, A good color name estimation is realized.
　次に、図１４及び図１５を参照して、推定部１２による色名の推定処理の他の例について説明する。以下に説明する例は、一の商品について、複数の商品画像が対応付けられており、商品画像の数と同数の色名候補が商品情報から抽出された場合に特に有効である。 Next, another example of color name estimation processing by the estimation unit 12 will be described with reference to FIGS. 14 and 15. The example described below is particularly effective when a plurality of product images are associated with one product and the same number of color name candidates as the number of product images are extracted from the product information.
　図１４（ａ）は、商品情報記憶部２１に記憶されたデータの例を示す図である。図１４（ａ）に示す例では、商品ＩＤ（Ｍ４）に対応付けて、商品情報Ｔ４及び商品画像Ｐ４１，Ｐ４２，Ｐ４３を記憶している。商品情報Ｔ４は、その内容として、「ダークブラウン、ダークブルー、ブラックの３色から選べます。」というユーザに提示するためのテキスト情報を含む。抽出部１１は、商品情報Ｔ４から、色名候補「ダークブラウン」、「ダークブルー」、「ブラック」を抽出する。 FIG. 14A is a diagram illustrating an example of data stored in the product information storage unit 21. In the example shown in FIG. 14A, product information T4 and product images P41, P42, and P43 are stored in association with the product ID (M4). The product information T4 includes, as its contents, text information for presenting to the user that “You can choose from three colors, dark brown, dark blue, and black.” The extraction unit 11 extracts color name candidates “dark brown”, “dark blue”, and “black” from the product information T4.
　商品ＩＤ（Ｍ４）の商品のように、一の商品について、複数の商品画像が対応付けられている場合、推定部１２は、複数の色名候補のうちの一の色名候補に対して設定された色範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出し、最もピクセル数が多い商品画像の色名が一の色名候補であると推定する。 When a plurality of product images are associated with one product like the product with the product ID (M4), the estimation unit 12 sets one color name candidate among the plurality of color name candidates. The number of pixels of the product image included in the determined color range is calculated for each of the plurality of product images, and it is estimated that the color name of the product image having the largest number of pixels is one color name candidate.
　具体的には、推定部１２は、一の色名候補「ダークブラウン」について設定された色範囲に含まれる商品画像のピクセルの数を、商品画像Ｐ４１，Ｐ４２，Ｐ４３のそれぞれについて算出する。同様に、推定部１２は、色名候補「ダークブルー」について設定された色範囲に含まれる商品画像のピクセルの数を、商品画像Ｐ４１，Ｐ４２，Ｐ４３のそれぞれについて算出し、色名候補「ブラック」について設定された色範囲に含まれる商品画像のピクセルの数を、商品画像Ｐ４１，Ｐ４２，Ｐ４３のそれぞれについて算出する。 Specifically, the estimation unit 12 calculates the number of product image pixels included in the color range set for one color name candidate “dark brown” for each of the product images P41, P42, and P43. Similarly, the estimation unit 12 calculates the number of product image pixels included in the color range set for the color name candidate “dark blue” for each of the product images P41, P42, and P43, and the color name candidate “black”. The number of pixels of the product image included in the color range set for “” is calculated for each of the product images P41, P42, and P43.
　図１５は、色名候補「ダークブラウン」、「ダークブルー」、「ブラック」のそれぞれについて算出された、商品画像Ｐ４１，Ｐ４２，Ｐ４３のそれぞれのピクセルが色範囲に含まれるピクセル数の例を示す図である。 FIG. 15 illustrates an example of the number of pixels in which the pixels of the product images P41, P42, and P43 are included in the color range, calculated for each of the color name candidates “dark brown”, “dark blue”, and “black”. FIG.
　図１５に示されるように、例えば、色名候補「ダークブラウン」の色範囲に含まれる商品画像Ｐ４１，Ｐ４２，Ｐ４３のピクセルの数はそれぞれ、２００，１８０，１７５である。従って、色名候補「ダークブラウン」の色範囲に含まれるピクセルを最も多く含む商品画像は商品画像Ｐ４１であるので、推定部１２は、商品画像Ｐ４１に表された商品の色名を「ダークブラウン」と推定する。 15, for example, the number of pixels of the product images P41, P42, and P43 included in the color range of the color name candidate “dark brown” is 200, 180, and 175, respectively. Therefore, since the product image including the most pixels included in the color range of the color name candidate “dark brown” is the product image P41, the estimation unit 12 assigns the color name of the product represented in the product image P41 to “dark brown”. ".
　また、色名候補「ダークブルー」の色範囲に含まれる商品画像Ｐ４１，Ｐ４２，Ｐ４３のピクセルの数はそれぞれ、１３０，２００，１４０である。従って、色名候補「ダークブルー」の色範囲に含まれるピクセルを最も多く含む商品画像は商品画像Ｐ４２であるので、推定部１２は、商品画像Ｐ４２に表された商品の色名を「ダークブルー」と推定する。 Further, the numbers of pixels of the product images P41, P42, and P43 included in the color range of the color name candidate “dark blue” are 130, 200, and 140, respectively. Accordingly, since the product image including the most pixels included in the color range of the color name candidate “dark blue” is the product image P42, the estimation unit 12 sets the color name of the product represented in the product image P42 to “dark blue”. ”.
　また、色名候補「ブラック」の色範囲に含まれる商品画像Ｐ４１，Ｐ４２，Ｐ４３のピクセルの数はそれぞれ、９０，８０，１３０である。従って、色名候補「ブラック」の色範囲に含まれるピクセルを最も多く含む商品画像は商品画像Ｐ４３であるので、推定部１２は、商品画像Ｐ４３に表された商品の色名を「ブラック」と推定する。 Also, the numbers of pixels of the product images P41, P42, and P43 included in the color range of the color name candidate “black” are 90, 80, and 130, respectively. Therefore, since the product image including the most pixels included in the color range of the color name candidate “black” is the product image P43, the estimation unit 12 sets the color name of the product represented in the product image P43 to “black”. presume.
　そして、図１４（ｂ）に示されるように、出力部１３は、推定された色名「ダークブラウン」、「ダークブルー」、「ブラック」のそれぞれを、商品情報記憶部２１の商品画像Ｐ４１，Ｐ４２，Ｐ４３に対応付けて記憶させる。 Then, as illustrated in FIG. 14B, the output unit 13 uses the estimated color names “dark brown”, “dark blue”, and “black” as product images P41, P42 and P43 are stored in association with each other.
　図１５に示した例において、商品画像ごとに各色名候補の色範囲に含まれるピクセル数を算出する処理により色名が推定されると、商品画像Ｐ４３に含まれるピクセルのうち、ダークブラウンに含まれるピクセル数が最も多いので、商品画像Ｐ４３の色名がダークブラウンと誤って推定される。しかしながら、以上説明した処理の例では、一の色名候補について、各商品画像の当該色名に相当するピクセル値のピクセルの数が算出され、最もピクセル数が多い商品画像の色名として、一の色名候補が推定されるので、抽出された複数の色名候補が近いピクセル値を有する場合であっても、精度良い色名の推定が可能となる。 In the example illustrated in FIG. 15, when the color name is estimated by the process of calculating the number of pixels included in the color range of each color name candidate for each product image, it is included in dark brown among the pixels included in the product image P43. Therefore, the product name P43 is erroneously estimated to be dark brown. However, in the processing example described above, for one color name candidate, the number of pixels of the pixel value corresponding to the color name of each product image is calculated, and the color name of the product image with the largest number of pixels is one. Therefore, even if the extracted color name candidates have similar pixel values, it is possible to estimate the color name with high accuracy.
　次に、図１６、図１７及び図１８を参照して、推定部１２による色名の推定処理のさらに別の例について説明する。図１６（ａ）は、商品情報記憶部２１に記憶されたデータの例を示す図である。図１６（ａ）に示す例では、商品ＩＤ（Ｍ５）に対応付けて、商品情報Ｔ５及び商品画像Ｐ５１，５２を記憶している。商品情報Ｔ５は、その内容として、「ブラックとダークブルーがあります。」というユーザに提示するためのテキスト情報を含む。抽出部１１は、商品情報Ｔ５から、色名候補「ブラック」、「ダークブルー」を抽出する。 Next, still another example of the color name estimation process performed by the estimation unit 12 will be described with reference to FIGS. 16, 17, and 18. FIG. 16A is a diagram illustrating an example of data stored in the product information storage unit 21. In the example shown in FIG. 16A, product information T5 and product images P51 and P52 are stored in association with the product ID (M5). The merchandise information T5 includes text information to be presented to the user, “There are black and dark blue.” The extraction unit 11 extracts color name candidates “black” and “dark blue” from the product information T5.
　まず、推定部１２は、商品画像Ｐ５１について、各色名候補に対して設定された色範囲に含まれる商品画像Ｐ５１のピクセルの数を色名候補ごとに算出する。図１７（ａ）は、商品画像Ｐ５１について、色名候補「ブラック」及び「ダークブルー」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。推定部１２は、算出されたピクセル数が最も多い「ブラック」を商品画像Ｐ５１に表された商品の色名として推定する。 First, for the product image P51, the estimation unit 12 calculates the number of pixels of the product image P51 included in the color range set for each color name candidate for each color name candidate. FIG. 17A is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P51. The estimation unit 12 estimates “black” having the largest number of calculated pixels as the color name of the product shown in the product image P51.
　同様に、推定部１２は、商品画像Ｐ５２について、各色名候補に対して設定された色範囲に含まれる商品画像Ｐ５２のピクセルの数を色名候補ごとに算出する。図１７（ｂ）は、商品画像Ｐ５２について、色名候補「ブラック」及び「ダークブルー」のそれぞれの色範囲に含まれるピクセル数の算出結果を示す図である。推定部１２は、算出されたピクセル数が最も多い「ブラック」を商品画像Ｐ５２に表された商品の色名として推定する。 Similarly, for the product image P52, the estimation unit 12 calculates the number of pixels of the product image P52 included in the color range set for each color name candidate for each color name candidate. FIG. 17B is a diagram illustrating a calculation result of the number of pixels included in the color ranges of the color name candidates “black” and “dark blue” for the product image P52. The estimation unit 12 estimates “black” having the largest number of calculated pixels as the color name of the product represented in the product image P52.
　そして、図１６（ｂ）に示されるように、出力部１３は、商品画像Ｐ５１について推定された色名「ブラック」，及び商品画像Ｐ５２について推定された色名「ブラック」を、商品情報記憶部２１の商品画像Ｐ５１，Ｐ５２に対応付けて記憶させる。図１６（ｂ）に示されるように、色名候補として複数の色名候補「ブラック」及び「ダークブルー」が抽出部１１により抽出されたにも関わらず、色名公報「ダークブルー」は、いずれの商品画像の商品の色名としても推定されていないので、色名の推定において誤りがある可能性が高い。このような場合に、推定部１２は、図１４及び図１５を参照して説明した例と同様に、複数の色名候補のうちの一の色名候補に対して設定された色範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出し、最もピクセル数が多い商品画像の色名が一の色名候補であると推定する。 Then, as illustrated in FIG. 16B, the output unit 13 outputs the color name “black” estimated for the product image P51 and the color name “black” estimated for the product image P52 to the product information storage unit. The product images P51 and P52 of 21 are stored in association with each other. As shown in FIG. 16B, the color name publication “Dark Blue” is the color name candidate “Dark Blue”, although a plurality of color name candidates “Black” and “Dark Blue” are extracted by the extraction unit 11. Since it is not estimated as the color name of the product of any product image, there is a high possibility that there is an error in the estimation of the color name. In such a case, the estimation unit 12 is included in the color range set for one color name candidate among the plurality of color name candidates, as in the example described with reference to FIGS. 14 and 15. The number of pixels of the product image to be obtained is calculated for each of the plurality of product images, and it is estimated that the color name of the product image having the largest number of pixels is one color name candidate.
　具体的には、推定部１２は、色名候補「ブラック」について設定された色範囲に含まれる商品画像のピクセルの数を、商品画像Ｐ５１，Ｐ５２のそれぞれについて算出する。同様に、推定部１２は、色名候補「ダークブルー」について設定された色範囲に含まれる商品画像のピクセルの数を、商品画像Ｐ５１，Ｐ５２のそれぞれについて算出する。図１７（ｃ）は、色名候補「ブラック」、「ダークブルー」のそれぞれについて算出された、商品画像Ｐ５１，Ｐ５２のそれぞれのピクセルが色範囲に含まれるピクセル数の例を示す図である。 Specifically, the estimation unit 12 calculates the number of product image pixels included in the color range set for the color name candidate “black” for each of the product images P51 and P52. Similarly, the estimation unit 12 calculates the number of product image pixels included in the color range set for the color name candidate “dark blue” for each of the product images P51 and P52. FIG. 17C is a diagram illustrating an example of the number of pixels that are calculated for each of the color name candidates “black” and “dark blue” and each pixel of the product images P51 and P52 is included in the color range.
　図１７（ｃ）に示されるように、色名候補「ブラック」の色範囲に含まれる商品画像Ｐ５１，Ｐ５２のピクセルの数はそれぞれ、１５０，１４０である。従って、色名候補「ブラック」の色範囲に含まれるピクセルを最も多く含む商品画像は商品画像Ｐ５１であるので、推定部１２は、商品画像Ｐ５１に表された商品の色名を「ブラック」と推定する。また、色名候補「ダークブルー」の色範囲に含まれる商品画像Ｐ５１，Ｐ５２のピクセルの数はそれぞれ、１００，１２０である。従って、色名候補「ダークブルー」の色範囲に含まれるピクセルを最も多く含む商品画像は商品画像Ｐ５２であるので、推定部１２は、商品画像Ｐ５２に表された商品の色名を「ダークブルー」と推定する。そして、図１６（ｃ）に示されるように、出力部１３は、推定された色名「ブラック」、「ダークブルー」のそれぞれを、商品情報記憶部２１の商品画像Ｐ５１，Ｐ５２に対応付けて記憶させる。 17C, the number of pixels of the product images P51 and P52 included in the color range of the color name candidate “black” is 150 and 140, respectively. Accordingly, since the product image including the most pixels included in the color range of the color name candidate “black” is the product image P51, the estimation unit 12 sets the color name of the product represented in the product image P51 to “black”. presume. Further, the numbers of pixels of the product images P51 and P52 included in the color range of the color name candidate “dark blue” are 100 and 120, respectively. Therefore, since the product image including the most pixels included in the color range of the color name candidate “dark blue” is the product image P52, the estimation unit 12 sets the color name of the product represented in the product image P52 to “dark blue”. ". Then, as illustrated in FIG. 16C, the output unit 13 associates the estimated color names “black” and “dark blue” with the product images P51 and P52 of the product information storage unit 21, respectively. Remember me.
　図１８は、推定部１２による色名の推定処理のさらに別の例の内容を示すフローチャートである。まず、抽出部１１は、商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を商品の色名候補として抽出する（Ｓ１１）。 FIG. 18 is a flowchart showing the contents of still another example of color name estimation processing by the estimation unit 12. First, the extraction unit 11 extracts a plurality of color text information representing color names as product color name candidates from the product information associated with the product image (S11).
　次に、推定部１２は、推定部１２は、色名候補に対して設定された色範囲に含まれる一の商品画像のピクセルの数を色名候補ごとに算出する（Ｓ１２）。そして、推定部１２は、算出されたピクセル数が最も多い色名候補を、一の商品画像に表された商品の色名として推定する（Ｓ１３）。 Next, the estimation unit 12 calculates, for each color name candidate, the number of pixels of one product image included in the color range set for the color name candidate (S12). Then, the estimation unit 12 estimates the color name candidate having the largest number of calculated pixels as the color name of the product represented in one product image (S13).
　続いて、推定部１２は、複数の色名候補のうち、いずれの商品画像の商品の色名として推定されなかった色名候補が存在するか否かを判定する（Ｓ１４）。いずれの商品画像の商品の色名として推定されなかった色名候補が存在すると判定された場合には、処理手順はステップＳ１５に進められる。一方、いずれの商品画像の商品の色名として推定されなかった色名候補が存在すると判定されなかった場合には、処理手順はステップＳ１７に進められる。 Subsequently, the estimation unit 12 determines whether there is a color name candidate that has not been estimated as the product color name of any product image among the plurality of color name candidates (S14). If it is determined that there is a color name candidate that has not been estimated as the product color name of any product image, the processing procedure proceeds to step S15. On the other hand, if it is not determined that there is a color name candidate that has not been estimated as the color name of the product of any product image, the processing procedure proceeds to step S17.
　ステップＳ１５において、推定部１２は、複数の色名候補のうちの一の色名候補に対して設定された色範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出する（Ｓ１５）。そして、推定部１２は、最もピクセル数が多い商品画像の色名が一の色名候補であると推定する（Ｓ１６）。出力部１３は、推定された色名を出力する（Ｓ１７）。 In step S15, the estimation unit 12 calculates the number of pixels of the product image included in the color range set for one color name candidate among the plurality of color name candidates for each of the plurality of product images. (S15). And the estimation part 12 estimates that the color name of the product image with the largest number of pixels is one color name candidate (S16). The output unit 13 outputs the estimated color name (S17).
　このように色名の推定処理では、一の商品画像について、各色名候補に対して設定された色範囲に含まれるピクセル数が最も多い色名候補を一の商品画像に表された商品の色名として推定し、全ての商品画像についての色名の推定を実施した結果、商品の色名として推定されなかった色名が存在する場合に、一の色名候補について、各商品画像の当該色名に相当するピクセル値のピクセルの数が算出され、最もピクセル数が多い商品画像の色名として、一の色名候補が推定される。これにより、一の商品についての複数の商品画像に複数の色名候補が対応付けられる場合において、複数の色名候補が近いピクセル値を有する場合であっても、精度良く商品の色名を推定できる。 In this way, in the color name estimation process, for one product image, the color name candidate having the largest number of pixels included in the color range set for each color name candidate is the color of the product represented in the one product image. As a result of estimating the color name for all product images, if there is a color name that has not been estimated as the color name of the product, the color of each product image for one color name candidate The number of pixels having the pixel value corresponding to the name is calculated, and one color name candidate is estimated as the color name of the product image having the largest number of pixels. As a result, when a plurality of color name candidates are associated with a plurality of product images of one product, the color name of the product is accurately estimated even when the plurality of color name candidates have similar pixel values. it can.
　以上、本発明をその実施形態に基づいて詳細に説明した。しかし、本発明は上記実施形態に限定されるものではない。本発明は、その要旨を逸脱しない範囲で様々な変形が可能である。 The present invention has been described in detail above based on the embodiments. However, the present invention is not limited to the above embodiment. The present invention can be variously modified without departing from the gist thereof.
　１…色推定装置、３…電子商取引サーバ、１１…抽出部、１２…推定部、１３…出力部、２１…商品情報記憶部、２２…色情報記憶部、３１…検索部、１ｄ…記憶媒体、１ｐ…色推定プログラム、ｍ１０…メインモジュール、ｍ１１…抽出モジュール、ｍ１２…推定モジュール、ｍ１３…出力モジュール。
  DESCRIPTION OF SYMBOLS 1 ... Color estimation apparatus, 3 ... Electronic commerce server, 11 ... Extraction part, 12 ... Estimation part, 13 ... Output part, 21 ... Merchandise information storage part, 22 ... Color information storage part, 31 ... Search part, 1d ... Storage medium DESCRIPTION OF SYMBOLS 1p ... Color estimation program, m10 ... Main module, m11 ... Extraction module, m12 ... Estimation module, m13 ... Output module.

Claims (16)

  1. 　商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を前記商品の色名候補として抽出する抽出手段と、
    　それぞれの前記色名候補に対して設定されたピクセル値と、前記商品画像に含まれるピクセルのピクセル値とに基づき、前記商品画像に表された商品の色を推定する推定手段と、
    　前記推定手段により推定された色を出力する出力手段と、
    　を備える色推定装置。     Extraction means for extracting a plurality of color text information representing a color name as color name candidates of the product from the product information associated with the product image representing the product;
    Estimating means for estimating a color of a product represented in the product image based on a pixel value set for each of the color name candidates and a pixel value of a pixel included in the product image;
    Output means for outputting the color estimated by the estimation means;
    A color estimation apparatus comprising:
  2. 　前記推定手段は、前記商品画像に表された商品の色を前記抽出手段により抽出された色名候補の中から推定する、
    　請求項１に記載の色推定装置。     The estimating means estimates the color of the product represented in the product image from the color name candidates extracted by the extracting means;
    The color estimation apparatus according to claim 1.
  3. 　前記推定手段は、前記色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる前記商品画像のピクセルの数を前記色名候補ごとに算出し、算出されたピクセル数が最も多い色名候補を、前記商品画像に表された商品の色として推定する、
    　請求項１または２に記載の色推定装置。     The estimation means calculates the number of pixels of the product image included in a color range that is a range of pixel values set for the color name candidate for each color name candidate, and the calculated number of pixels is the largest. A large number of color name candidates are estimated as the color of the product represented in the product image.
    The color estimation apparatus according to claim 1 or 2.
  4. 　前記推定手段は、前記商品画像に含まれる第１のピクセルより前記商品画像の中央部に近い位置にある第２のピクセルに対して、前記第１のピクセルより大きい重み付けをして、前記色名候補ごとの前記ピクセル数の算出を行う、
    　請求項３に記載の色推定装置。     The estimation means weights the second pixel located closer to the center of the product image than the first pixel included in the product image to a greater weight than the first pixel, and the color name Calculating the number of pixels for each candidate;
    The color estimation apparatus according to claim 3.
  5. 　前記推定手段は、前記色名候補に対して設定されたピクセル値に基づき、前記色名候補ごとの色範囲を重複が生じないように設定する、
    　請求項３または４に記載の色推定装置。     The estimation means sets a color range for each color name candidate so as not to overlap based on a pixel value set for the color name candidate.
    The color estimation apparatus according to claim 3 or 4.
  6. 　前記推定手段は、前記商品画像のピクセルをピクセル値によりクラスタリングし、グループの中央値が前記色名候補の色範囲に含まれる場合に、該グループに含まれるピクセルの数を該色名候補の色範囲に含まれるピクセル数として、前記色名候補ごとのピクセル数の算出を行う、
    　請求項３～５のいずれか一項に記載の色推定装置。     The estimation unit clusters the pixels of the product image by pixel values, and when the median value of the group is included in the color range of the color name candidate, the number of pixels included in the group is determined as the color of the color name candidate. The number of pixels for each color name candidate is calculated as the number of pixels included in the range.
    The color estimation apparatus according to any one of claims 3 to 5.
  7. 　前記推定手段は、算出されたピクセル数が最も多い色名候補、及び該色名候補と算出されたピクセル数の差が所定数以下のピクセル数の色名候補を、前記商品画像に表された複数の商品のそれぞれの色として推定する、
    　請求項３～６のいずれか一項に記載の色推定装置。     The estimation means displays the color name candidate having the largest number of calculated pixels, and the color name candidate having the number of pixels having a difference between the color name candidate and the calculated number of pixels equal to or less than a predetermined number in the product image. Estimate as each color of multiple products,
    The color estimation apparatus according to any one of claims 3 to 6.
  8. 　前記推定手段は、一の商品について、複数の商品画像が対応付けられている場合、一の商品画像ごとに色を推定する、
    　請求項３～７のいずれか一項に記載の色推定装置。     The estimation means estimates a color for each product image when a plurality of product images are associated with one product.
    The color estimation apparatus according to any one of claims 3 to 7.
  9. 　前記推定手段は、一の商品について、複数の商品画像が対応付けられている場合、前記複数の色名候補のうちの一の前記色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出し、最もピクセル数が多い商品画像の色が前記一の色名候補であると推定する、
    　請求項１または２に記載の色推定装置。     The estimation means, when a plurality of product images are associated with one product, a color that is a range of pixel values set for one of the plurality of color name candidates The number of product image pixels included in the range is calculated for each of a plurality of product images, and the color of the product image with the largest number of pixels is estimated as the one color name candidate.
    The color estimation apparatus according to claim 1 or 2.
  10. 　前記推定手段は、一の商品について、複数の商品画像が対応付けられている場合、前記色名候補に対して設定されたピクセル値の範囲である色範囲に含まれる一の商品画像のピクセルの数を前記色名候補ごとに算出し、算出されたピクセル数が最も多い色名候補を、前記一の商品画像に表された商品の色として推定し、
    　複数の色名候補のうち、いずれの商品画像の商品の色として推定されなかった色名候補が存在する場合に、前記複数の色名候補のうちの一の色名候補に対して設定されたピクセル値の範囲に含まれる商品画像のピクセルの数を、複数の商品画像のそれぞれについて算出し、最もピクセル数が多い商品画像の色が前記一の色名候補であると推定する、
    　請求項１または２に記載の色推定装置。     In the case where a plurality of product images are associated with one product, the estimating means determines the pixel of one product image included in a color range that is a pixel value range set for the color name candidate. The number is calculated for each color name candidate, the color name candidate having the largest number of calculated pixels is estimated as the color of the product represented in the one product image,
    When there is a color name candidate that has not been estimated as the product color of any product image among the plurality of color name candidates, it is set for one color name candidate among the plurality of color name candidates Calculating the number of pixels of the product image included in the pixel value range for each of the plurality of product images, and estimating that the color of the product image having the largest number of pixels is the one color name candidate.
    The color estimation apparatus according to claim 1 or 2.
  11. 　前記抽出手段は、前記商品情報に商品の色の指定をユーザから受け付けるための欄が含まれる場合、前記欄の色テキスト情報を色名候補として抽出する、
    　請求項１～１０のいずれか一項に記載の色推定装置。     The extraction means, when the product information includes a column for accepting designation of a product color from a user, extracts the color text information of the column as a color name candidate,
    The color estimation apparatus according to any one of claims 1 to 10.
  12. 　出力手段は、前記商品情報と前記商品画像とを対応付けて記憶している商品情報記憶手段の商品画像に、推定された色を対応付けて記憶させる、
    　請求項１～１１のいずれか一項に記載の色推定装置。     The output unit stores the estimated color in association with the product image of the product information storage unit that stores the product information and the product image in association with each other.
    The color estimation apparatus according to any one of claims 1 to 11.
  13. 　商品の色の指定を含むユーザからの検索要求に応じて、前記商品情報記憶手段を参照して、指定された色が対応づけられた商品画像を含む検索結果を該ユーザに返信する検索手段をさらに備える請求項１２に記載の色推定装置。 Search means for returning a search result including a product image associated with the designated color to the user with reference to the product information storage means in response to a search request from the user including designation of the product color The color estimation device according to claim 12, further comprising:
  14. 　前記商品情報記憶手段において、一の商品について複数の商品画像が対応付けられており、前記複数の商品画像のうちの一の商品画像が商品を提示するための商品ページにおいて該商品を示すための代表画像として設定されている場合に、前記検索手段は、前記代表画像に対応づけられた色と異なる色の指定を含むユーザからの検索要求を受け付けた場合に、指定された色が対応づけられた商品画像を代表画像として含む検索結果を該ユーザに返信する、
    　請求項１３に記載の色推定装置。     In the product information storage unit, a plurality of product images are associated with one product, and one product image of the plurality of product images indicates the product on a product page for presenting the product. When set as a representative image, the search means associates the specified color when receiving a search request from a user including a color specification different from the color associated with the representative image. A search result including the selected product image as a representative image is returned to the user.
    The color estimation apparatus according to claim 13.
  15. 　コンピュータにより実行される色推定方法であって、
    　商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を前記商品の色名候補として抽出する抽出ステップと、
    　それぞれの前記色名候補に対して設定されたピクセル値と、前記商品画像に含まれるピクセルのピクセル値とに基づき、前記商品画像に表された商品の色を推定する推定ステップと、
    　前記推定ステップにおいて推定された色を出力する出力ステップと、
    　を有する色推定方法。     A color estimation method executed by a computer,
    An extraction step of extracting a plurality of color text information representing a color name as color name candidates of the product from the product information associated with the product image representing the product;
    An estimation step of estimating a color of a product represented in the product image based on a pixel value set for each of the color name candidates and a pixel value of a pixel included in the product image;
    An output step of outputting the color estimated in the estimation step;
    A color estimation method comprising:
  16. 　コンピュータに、
    　商品が表された商品画像に関連付けられた商品情報から、色名を表す複数の色テキスト情報を前記商品の色名候補として抽出する抽出機能と、
    　それぞれの前記色名候補に対して設定されたピクセル値と、前記商品画像に含まれるピクセルのピクセル値とに基づき、前記商品画像に表された商品の色を推定する推定機能と、
    　前記推定機能により推定された色を出力する出力機能と、
    　を実現させる色推定プログラム。
     
     
          On the computer,
    An extraction function for extracting a plurality of color text information representing a color name as color name candidates of the product from the product information associated with the product image in which the product is represented;
    An estimation function for estimating a color of a product represented in the product image based on a pixel value set for each of the color name candidates and a pixel value of a pixel included in the product image;
    An output function for outputting the color estimated by the estimation function;
    Color estimation program that realizes


PCT/JP2014/059303 2014-03-28 2014-03-28 Color estimation device, color estimation method, and color estimation program WO2015145766A1 (en)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2014527102A JP5687806B1 (en) 2014-03-28 2014-03-28 Color estimation apparatus, color estimation method, and color estimation program
US15/022,243 US10692133B2 (en) 2014-03-28 2014-03-28 Color estimation device, color estimation method, and color estimation program
PCT/JP2014/059303 WO2015145766A1 (en) 2014-03-28 2014-03-28 Color estimation device, color estimation method, and color estimation program

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2014/059303 WO2015145766A1 (en) 2014-03-28 2014-03-28 Color estimation device, color estimation method, and color estimation program

Publications (1)

Publication Number Publication Date
WO2015145766A1 true WO2015145766A1 (en) 2015-10-01

Family

ID=52823256

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2014/059303 WO2015145766A1 (en) 2014-03-28 2014-03-28 Color estimation device, color estimation method, and color estimation program

Country Status (3)

Country Link
US (1) US10692133B2 (en)
JP (1) JP5687806B1 (en)
WO (1) WO2015145766A1 (en)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6028130B1 (en) * 2016-02-09 2016-11-16 楽天株式会社 Color classification apparatus, color classification method, program, and non-transitory computer-readable information recording medium
US9785649B1 (en) 2014-09-02 2017-10-10 Amazon Technologies, Inc. Hue-based color naming for an image
US9792303B2 (en) 2014-06-26 2017-10-17 Amazon Technologies, Inc. Identifying data from keyword searches of color palettes and keyword trends
US9836856B2 (en) 2014-06-26 2017-12-05 Amazon Technologies, Inc. Color name generation from images and color palettes
US9898487B2 (en) 2014-06-26 2018-02-20 Amazon Technologies, Inc. Determining color names from keyword searches of color palettes
US9916613B1 (en) 2014-06-26 2018-03-13 Amazon Technologies, Inc. Automatic color palette based recommendations for affiliated colors
US9922050B2 (en) 2014-06-26 2018-03-20 Amazon Technologies, Inc. Identifying data from keyword searches of color palettes and color palette trends
US9996579B2 (en) 2014-06-26 2018-06-12 Amazon Technologies, Inc. Fast color searching
US10073860B2 (en) 2014-06-26 2018-09-11 Amazon Technologies, Inc. Generating visualizations from keyword searches of color palettes
US10120880B2 (en) 2014-06-26 2018-11-06 Amazon Technologies, Inc. Automatic image-based recommendations using a color palette
US10169803B2 (en) 2014-06-26 2019-01-01 Amazon Technologies, Inc. Color based social networking recommendations
US10186054B2 (en) 2014-06-26 2019-01-22 Amazon Technologies, Inc. Automatic image-based recommendations using a color palette
US10223427B1 (en) 2014-06-26 2019-03-05 Amazon Technologies, Inc. Building a palette of colors based on human color preferences
US10235389B2 (en) 2014-06-26 2019-03-19 Amazon Technologies, Inc. Identifying data from keyword searches of color palettes
US10255295B2 (en) 2014-06-26 2019-04-09 Amazon Technologies, Inc. Automatic color validation of image metadata
US10402917B2 (en) 2014-06-26 2019-09-03 Amazon Technologies, Inc. Color-related social networking recommendations using affiliated colors
US10430857B1 (en) * 2014-08-01 2019-10-01 Amazon Technologies, Inc. Color name based search
US10691744B2 (en) 2014-06-26 2020-06-23 Amazon Technologies, Inc. Determining affiliated colors from keyword searches of color palettes

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10706450B1 (en) 2018-02-14 2020-07-07 Amazon Technologies, Inc. Artificial intelligence system for generating intent-aware recommendations
US11679014B2 (en) 2018-10-30 2023-06-20 Boston Scientific Scimed, Inc. Devices and methods for treatment of body lumens

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002259401A (en) * 2001-02-27 2002-09-13 Olympus Optical Co Ltd Commodity search system, commodity search method and recording medium
JP2010055391A (en) * 2008-08-28 2010-03-11 Fujifilm Corp Merchandise analyzing device, merchandise offering device, and merchandise analysis method and program
WO2011045920A1 (en) * 2009-10-16 2011-04-21 日本電気株式会社 Color analysis device, color analysis method, and color analysis program
JP2011520203A (en) * 2008-05-09 2011-07-14 エルティーユー テクノロジーズ エスエーエス Color matching tool box

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8532372B2 (en) * 2010-08-19 2013-09-10 Thomas Youngman System and method for matching color swatches
US20130300761A1 (en) * 2010-11-12 2013-11-14 Colormodules Inc. Method and system for color matching and color recommendation
WO2013021889A1 (en) * 2011-08-05 2013-02-14 楽天株式会社 Color name determination device, color name determination method, information recording medium, and program
US9047633B2 (en) * 2012-02-07 2015-06-02 Zencolor Corporation System and method for identifying, searching and matching products based on color
US9330340B1 (en) * 2014-05-06 2016-05-03 Google Inc. Noise estimation for images using polynomial relationship for pixel values of image features

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002259401A (en) * 2001-02-27 2002-09-13 Olympus Optical Co Ltd Commodity search system, commodity search method and recording medium
JP2011520203A (en) * 2008-05-09 2011-07-14 エルティーユー テクノロジーズ エスエーエス Color matching tool box
JP2010055391A (en) * 2008-08-28 2010-03-11 Fujifilm Corp Merchandise analyzing device, merchandise offering device, and merchandise analysis method and program
WO2011045920A1 (en) * 2009-10-16 2011-04-21 日本電気株式会社 Color analysis device, color analysis method, and color analysis program

Cited By (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10073860B2 (en) 2014-06-26 2018-09-11 Amazon Technologies, Inc. Generating visualizations from keyword searches of color palettes
US11216861B2 (en) 2014-06-26 2022-01-04 Amason Technologies, Inc. Color based social networking recommendations
US10120880B2 (en) 2014-06-26 2018-11-06 Amazon Technologies, Inc. Automatic image-based recommendations using a color palette
US9792303B2 (en) 2014-06-26 2017-10-17 Amazon Technologies, Inc. Identifying data from keyword searches of color palettes and keyword trends
US9836856B2 (en) 2014-06-26 2017-12-05 Amazon Technologies, Inc. Color name generation from images and color palettes
US9898487B2 (en) 2014-06-26 2018-02-20 Amazon Technologies, Inc. Determining color names from keyword searches of color palettes
US9916613B1 (en) 2014-06-26 2018-03-13 Amazon Technologies, Inc. Automatic color palette based recommendations for affiliated colors
US9922050B2 (en) 2014-06-26 2018-03-20 Amazon Technologies, Inc. Identifying data from keyword searches of color palettes and color palette trends
US9996579B2 (en) 2014-06-26 2018-06-12 Amazon Technologies, Inc. Fast color searching
US10169803B2 (en) 2014-06-26 2019-01-01 Amazon Technologies, Inc. Color based social networking recommendations
US10691744B2 (en) 2014-06-26 2020-06-23 Amazon Technologies, Inc. Determining affiliated colors from keyword searches of color palettes
US10402917B2 (en) 2014-06-26 2019-09-03 Amazon Technologies, Inc. Color-related social networking recommendations using affiliated colors
US10049466B2 (en) 2014-06-26 2018-08-14 Amazon Technologies, Inc. Color name generation from images and color palettes
US10186054B2 (en) 2014-06-26 2019-01-22 Amazon Technologies, Inc. Automatic image-based recommendations using a color palette
US10223427B1 (en) 2014-06-26 2019-03-05 Amazon Technologies, Inc. Building a palette of colors based on human color preferences
US10235389B2 (en) 2014-06-26 2019-03-19 Amazon Technologies, Inc. Identifying data from keyword searches of color palettes
US10242396B2 (en) 2014-06-26 2019-03-26 Amazon Technologies, Inc. Automatic color palette based recommendations for affiliated colors
US10255295B2 (en) 2014-06-26 2019-04-09 Amazon Technologies, Inc. Automatic color validation of image metadata
US10430857B1 (en) * 2014-08-01 2019-10-01 Amazon Technologies, Inc. Color name based search
US9785649B1 (en) 2014-09-02 2017-10-10 Amazon Technologies, Inc. Hue-based color naming for an image
US10831819B2 (en) 2014-09-02 2020-11-10 Amazon Technologies, Inc. Hue-based color naming for an image
WO2017138088A1 (en) * 2016-02-09 2017-08-17 楽天株式会社 Color classification device, color classification method, program, and non-transient computer-readable information recording medium
JP6028130B1 (en) * 2016-02-09 2016-11-16 楽天株式会社 Color classification apparatus, color classification method, program, and non-transitory computer-readable information recording medium

Also Published As

Publication number Publication date
US20160232601A1 (en) 2016-08-11
JPWO2015145766A1 (en) 2017-04-13
US10692133B2 (en) 2020-06-23
JP5687806B1 (en) 2015-03-25

Similar Documents

Publication Publication Date Title
JP5687806B1 (en) Color estimation apparatus, color estimation method, and color estimation program
JP7147921B2 (en) Image processing device, image processing method and program
US9865063B2 (en) Method and system for image feature extraction
US9208379B2 (en) Image processing apparatus, image processing method, image processing system, and storage medium storing program
US10891476B2 (en) Method, system, and neural network for identifying direction of a document
US9311532B2 (en) Article estimating system, article estimating method, and article estimating program
CN105518668A (en) Content based image retrieval
JP2016194834A (en) Conformity determination method, conformity determination program, and conformity determination system
CN107680112B (en) Image registration method
CN112330383A (en) Apparatus and method for visual element-based item recommendation
CN110807110A (en) Image searching method and device combining local and global features and electronic equipment
US8249363B2 (en) Image comparison system and method
CN112991555B (en) Data display method, device, equipment and storage medium
JP2016021101A (en) Design change management support device, and method
JP6822484B2 (en) Information processing equipment, information processing methods and programs
WO2014061285A1 (en) Corpus generating device, corpus generating method, and corpus generating program
US20160196259A1 (en) Information providing server and method of controlling information providing server
JP2020123319A (en) Method, apparatus, electronic device, computer-readable storage medium, and computer program for image-based data processing
US9245351B2 (en) Color evaluation apparatus, color evaluation method and computer program
US20210366018A1 (en) Server and method for avoiding registered materials
CN111563181B (en) Digital image file query method, device and readable storage medium
US11494553B2 (en) Document creation assistance apparatus and non-transitory computer readable medium
JP2022049110A (en) Image processing device, image processing system, and program
JP7138264B1 (en) Information processing device, information processing method, information processing system, and program
EP4195135A1 (en) Information processing device, information processing method, information processing system, and program

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2014527102

Country of ref document: JP

Kind code of ref document: A

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14887341

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15022243

Country of ref document: US

NENP Non-entry into the national phase
122 Ep: pct application non-entry in european phase

Ref document number: 14887341

Country of ref document: EP

Kind code of ref document: A1